Kevin C. Kelly, PharmD, BCPS

The US Department of Veterans Affairs (VA) annually treats around 450,000 veterans with cancer and diagnoses an additional 56,000.^1,2 Oral multikinase inhibitors (MKIs) are widely used as targeted therapies for many different malignancies. Despite the ease of oral administration, these agents are often accompanied by significant adverse effects (AEs) and drug-drug interactions.^3,4 Common AEs include hypertension, cutaneous reactions, gastrointestinal disturbances, proteinuria, and fatigue. Some serious outcomes that may occur are myocardial infarction, thrombosis, nephrotic syndrome, hemorrhage, hepatotoxicity, and gastrointestinal events.^5,6 Due to poor tolerability of these AEs, dose reductions, frequent therapy holds, and discontinuation of therapy may occur.

The US Food and Drug Administration recognizes dosing challenges with novel therapies and has created the Oncology Center of Excellence (OCE) Project Optimus initiative to reform dose optimization in oncology drug development. The initiative aims to shift the focus from establishing dose regimens based on the maximum tolerated doses of cytotoxic chemotherapeutics to an emphasis on maximum efficacy, safety, and tolerability, which better reflect real-world dosing.^7,8

MKIs can be challenging to manage because of the frequent toxicity-related dose reductions, interruptions, and discontinuations. In a multicenter retrospective study, Schnadig et al investigated dosing characteristics of first-line sunitinib for advanced renal cell carcinoma (RCC) and found that, among 114 patients who experienced AEs while taking sunitinib, 39.5% had dose reductions, 5.3% delayed therapy, 18.4% required additional supportive medications, and 22.8% discontinued sunitinib.⁹ Overall survival and median progression-free survival of these patients were lower than reported by Motzer et al in a phase 3 clinical trial.¹⁰ Schnadig et al concluded that patients treated with sunitinib for RCC in the community setting required more frequent dose reductions and had less time on therapy compared with patients in clinical trials, which ultimately impacted clinical outcomes.⁹

At the VA North Texas Health Care System (VANTHCS), patients with cancer have difficulty tolerating MKIs and often require dose alterations and/or discontinuation because of drug intolerance rather than discontinuation due to progression. Frequent dose adjustments for toxicity management can place more strain on patients and health care resources because of additional appointments, clinician time, and emergency department visits. Escalating drug costs can also cause concern when prescription doses are unused or changed frequently.

To capture and quantify prescribing practices and dose adjustments, this study evaluated the tolerability of MKIs at VANTHCS. This analysis may also guide clinicians in the selection of starting doses as well as dose titration expectations to optimize MKI therapy.

METHODS

This single-center, retrospective chart review analyzed patients receiving oral oncology MKIs for various malignancies at VANTHCS between January 1, 2014, and October 31, 2024. Participants included adults aged ≥ 18 years with a prescription for axitinib, cabozantinib, lenvatinib, pazopanib, regorafenib, sorafenib, or sunitinib initiated by the hematology/oncology service at VANTHCS. Patients were included if they had follow-up documentation with the hematology/oncology service and/or other VANTHCS clinicians outlining their course of therapy after MKI initiation. Patients were excluded if they did not have sufficient follow-up documentation (eg, transferred care to a non-VA health care practitioner [HCP], moved to another VA health care system), were enrolled in clinical trials, or were prescribed an MKI from a Care in the Community (CITC) prescriber. Electronic health record review and data collection were performed using the VA Computerized Patient Record System and Research Electronic Data Capture. Data were collected from the time of initiation to cessation of therapy and included information regarding therapy changes, progressive disease, and date of death, when available. Data collected included age, sex, race, comorbidities, date of death, type of malignancy and subtypes, cancer stage, MKI used (ie, drug, dose, frequency, schedule, and indication), dates of medication changes (ie, start, adjustment, hold, discontinuation), concurrent antineoplastic treatments, and AEs documented at times of dose change or interruption.

The primary outcome was MKI tolerance determined using relative dose intensity (RDI) and mean and median time on therapy. Two methods are used to calculate RDI that vary in how they approach time on therapy as outlined in Hawn et al.¹¹ This study used method 2, which accounts for holds in therapy by comparing the actual duration of treatment with the duration expected according to treatment protocol. Method 1 compares the prescribed dose with the administered dose and does not adjust for holds.¹¹ Using method 2, the RDI in this study was calculated by dividing the total actual dose given by the total indicated dose for the malignancy being treated, which accounts for duration of treatment.

The total actual dose was the strength, frequency, and days on therapy for each time frame of treatment multiplied together. This method accounted for all dose adjustments and time periods of treatment holds, including patient self-adjustments, prescriber-directed adjustments, and nonadherence determined by HCP documentation and/or prescription data. Similarly, the indicated total dose was calculated by multiplying the indicated strength, frequency, and all days that treatment should have occurred (time from start to finish). Indicated doses were derived from the prescribing information for each malignancy with the exception of sunitinib, for which the off-label dose of 37.5 mg daily was considered a full dose.^12,13 The total indicated dose for axitinib was calculated by considering the dose escalation schedule from the prescribing information.

Patients who required dose reductions due to renal/hepatic impairments or drug-drug interactions had their total indicated dose calculated using dose adjustments listed in the prescribing information. The mean RDI for each MKI agent was calculated by averaging the RDI for each prescription. The overall combined mean RDI included the means of all the MKIs reviewed to avoid skewing the results toward an MKI with more prescriptions. RDIs were also calculated for each cancer type for each agent. Additional descriptive secondary outcomes included rates of AEs and adjustments in doses.

RESULTS

Electronic data extraction identified 278 patients with 366 MKI prescriptions, of which 108 veterans with 158 MKI prescriptions were excluded. The top reason for exclusion was patients managed through CITC. Ultimately, 170 veterans with 208 MKI prescriptions managed by the VANTHCS hematology/oncology clinic were included (Table 1). Among patients receiving MKIs, the mean age was 72.7 years, 98% were male, and 99% had metastatic disease.

The overall combined mean MKI RDI was 67.5% using method 2 and ranged from 85.5% for sunitinib to 49.0% for sorafenib (Figure 1). Additional information regarding mean and median RDIs using method 2 is shown in Figure 1 and further subdivided by cancer type in Table 2. Median RDIs overall were similar to mean RDIs for most agents. Figure 2 indicates the mean and median time on therapy, reflecting time on therapy excluding days therapy was held. The overall combined mean and median days on therapy for all MKIs were 155 days and 95 days, respectively. Mean time on therapy depended on the agent used and ranged from 35 days (regorafenib) to 237 days (cabozantinib).

Of 208 MKI prescriptions, 127 (61.1%) were initiated at a reduced dose due to baseline concerns for tolerance such as performance status, frailty, and prior intolerance of other treatments. Eighty-one prescriptions (38.9%) were initiated at their indicated doses. Ninety prescriptions (43.3%) required dose reductions during treatment. Some MKI prescriptions had multiple dose increases and decreases, which is why RDI more accurately reflects dose adjustments. A total of 376 AEs that contributed to a dose adjustment, hold, or discontinuation occurred across all MKI prescriptions. The most common AEs were 82 failure-to-thrive events (21.8%) (fatigue, malaise, loss of appetite, reduced mobility, global decline), 79 gastrointestinal events (21.0%) (nausea, vomiting, diarrhea, abdominal pain), 62 dermatologic events (16.5%) (rash, hand-foot skin reactions, allergic response), 61 hepatic dysfunction events (16.2%) (liver enzyme elevations, hyperbilirubinemia), 40 cardiovascular events (10.6%) (hypertension, heart failure exacerbations, edema), and 33 renal dysfunction events (8.8%) (acute kidney injury, proteinuria) (Appendix 1).

DISCUSSION

The mean RDI of MKI prescriptions used in the veteran population at VANTHCS was about two-thirds of the indicated dose. These results indicate that most veterans required dose reductions and/or holds due to concerns over initial tolerance/performance status, worsening clinical condition, and/or intolerable AEs attributed to treatment. A retrospective study conducted by Denduluri et al suggested that an RDI of < 85% is a clinically meaningful reduction for traditional chemotherapy based on previous literature.¹⁴ However, it is less clear what RDI should be expected specifically for MKIs in real-world populations. The MKI phase 3 approval trials in RCC for axitinib, lenvatinib, and sunitinib found median RDIs of 89.4%, 69.6% to 70.4%, and 83.9%, respectively. Each trial cited dose reductions most commonly as the result of treatment-related AEs.^15,16

Studies on the impact of RDIs on survival outcomes found that higher RDIs may improve overall and progression-free survival. Retrospective studies inspecting lenvatinib in hepatocellular carcinoma (HCC) indicated that an RDI > 70% in the initial 4 weeks resulted in favorable survival outcomes.¹⁷ Similarly, a retrospective study investigating sunitinib in RCC found that an RDI > 60% conferred favorable survival outcomes.¹⁸ Alghamdi et al noted that patients taking sorafenib for HCC who had RDI > 50% had a favorable trend in survival characteristics. Interestingly, the study found an RDI of 50% to 75% appeared to have better survival than an RDI > 75%.¹⁹ The authors of these studies hypothesized that additional dose reductions allowed for longer total time on therapy due to improved tolerability.^17-19

This analysis found that the RDIs for most MKI agents at VANTHCS were < 85% and lower than the RDIs found in other review articles and phase 3 trials, with the exceptions of pazopanib in thyroid cancer and sunitinib in gastrointestinal stromal tumor (GIST), thyroid cancer, and neuroendocrine cancer. The reasons for the lower RDIs in this study are likely multifactorial, reflecting patient population characteristics, off-label dosing practices, and HCP experiences with these agents. Many veterans have chronic comorbidities that could contribute to reduced performance status and ability to tolerate these therapies. Despite attempts to preemptively reduce doses for patients and account for potential impaired tolerance, there were patients who required further dose reductions in our study.

Failure to thrive was the most common AE leading to dose adjustment or discontinuation, which illustrates the extensive effects these agents have on patient functioning in a real-world population. Notably though, the RDI for sunitinib was higher in this population because about half of patients were dosed using the off-label recommendation, whereas the prescribing information recommends a more intensive 6-week dosing cycle for certain cancer types.^12,13,20 Sorafenib was also often dose-adjusted based on a pharmacokinetic study of sorafenib in renal/hepatic dysfunction, and the RDI likely reflects the off-label prescribing pattern.²¹

Patients with thyroid cancer were found to have higher RDIs compared with those receiving the same agents for other cancer types. Improved tolerability of MKIs in thyroid cancer may be due to a generally more tolerable disease course. Thyroid cancer is the most common cancer in individuals aged < 40 years, a population that is often more robust with fewer comorbidities. Moreover, the 5-year relative survival rate for thyroid cancer remains > 98%.²² This rate is in contrast to those for other cancer types such as HCC, with a 5-year relative survival rate of only 15%.²³

It is challenging to compare the mean and median times on therapy found in this study with those in current literature, as this review included multiple different cancer types for each agent. However, the numbers are generally lower than durations of therapy found across the different disease states and further emphasize the difficulty in tolerating MKIs in the VANTHCS population. Regorafenib had a short duration of time on therapy, which highlights the importance of trials like ReDOS and initiatives such as OCE Project Optimus in helping improve tolerance.^7,8,24

Comparing our results with other studies proved challenging because the RDI calculation methods were not specified. Calculating RDIs in this study using method 1, which does not account for holds, resulted in higher RDIs (Appendix 2). Using method 1, all MKIs had RDIs < 85%, except for pazopanib in thyroid cancer (100%) and RCC (87.9%), and sunitinib in GIST (93.6%), thyroid cancer (100%), and neuroendocrine cancer (93.7%). Notably, using method 1 increased the RDI for pazopanib in neuroendocrine cancer from 5.4% to 50.0%. The low RDI was attributed to a single veteran with a long hold duration, which demonstrates the discrepancy that can occur between the 2 methods.

Limitations

The retrospective design, lack of survival outcomes, and difficulty comparing results with other literature were limitations of this study. Because survival outcomes were not evaluated, future research should seek to investigate how RDIs and dose adjustments made among MKIs can affect survival outcomes in real-world populations. This veteran population with cancer often had multiple chronic comorbidities, which may have contributed to difficulty tolerating MKIs and could have impacted results. Disease-related factors may have influenced the poor tolerance of the MKIs and were not specifically accounted for. Adjustment for comorbidities was not possible because of discrepancies and/or incomplete diagnosis codes and Eastern Cooperative Oncology Group performance status scores documented in patient charts. Therefore, we decided not to report these findings due to potential inaccuracies.

CONCLUSIONS

Results of this study demonstrate that oncology MKI agents used at VANTHCS were difficult for patients to tolerate, leading to suboptimal dosing compared with indicated doses established in clinical trials and prescribing information. Clinicians may use these data to help guide clinical decision-making whenever initiating and managing MKI agents in this population. These findings reinforce that MKI agents are often difficult to tolerate in real-world practice, and indicated doses are often not achieved. Further studies should aim to investigate the effect that various RDIs have on overall survival. Further investigation into different dosing schemes for MKIs to improve tolerability and longer-term use may also prove beneficial.

This analysis may help guide clinicians to carefully approach dosing MKI agents in the veteran population. Given the RDI and AEs, more clinicians may consider starting at lower than indicated doses with the goal to titrate up as tolerated. Additionally, the results highlight the importance of considering palliative care consults and ensuring appropriate supportive care agents are preemptively engaged and adjusted as needed. Approaching dosing and titrations cautiously may help reduce the burden of management on the health care system.

The US Department of Veterans Affairs (VA) annually treats around 450,000 veterans with cancer and diagnoses an additional 56,000.^1,2 Oral multikinase inhibitors (MKIs) are widely used as targeted therapies for many different malignancies. Despite the ease of oral administration, these agents are often accompanied by significant adverse effects (AEs) and drug-drug interactions.^3,4 Common AEs include hypertension, cutaneous reactions, gastrointestinal disturbances, proteinuria, and fatigue. Some serious outcomes that may occur are myocardial infarction, thrombosis, nephrotic syndrome, hemorrhage, hepatotoxicity, and gastrointestinal events.^5,6 Due to poor tolerability of these AEs, dose reductions, frequent therapy holds, and discontinuation of therapy may occur.

The US Food and Drug Administration recognizes dosing challenges with novel therapies and has created the Oncology Center of Excellence (OCE) Project Optimus initiative to reform dose optimization in oncology drug development. The initiative aims to shift the focus from establishing dose regimens based on the maximum tolerated doses of cytotoxic chemotherapeutics to an emphasis on maximum efficacy, safety, and tolerability, which better reflect real-world dosing.^7,8

MKIs can be challenging to manage because of the frequent toxicity-related dose reductions, interruptions, and discontinuations. In a multicenter retrospective study, Schnadig et al investigated dosing characteristics of first-line sunitinib for advanced renal cell carcinoma (RCC) and found that, among 114 patients who experienced AEs while taking sunitinib, 39.5% had dose reductions, 5.3% delayed therapy, 18.4% required additional supportive medications, and 22.8% discontinued sunitinib.⁹ Overall survival and median progression-free survival of these patients were lower than reported by Motzer et al in a phase 3 clinical trial.¹⁰ Schnadig et al concluded that patients treated with sunitinib for RCC in the community setting required more frequent dose reductions and had less time on therapy compared with patients in clinical trials, which ultimately impacted clinical outcomes.⁹

At the VA North Texas Health Care System (VANTHCS), patients with cancer have difficulty tolerating MKIs and often require dose alterations and/or discontinuation because of drug intolerance rather than discontinuation due to progression. Frequent dose adjustments for toxicity management can place more strain on patients and health care resources because of additional appointments, clinician time, and emergency department visits. Escalating drug costs can also cause concern when prescription doses are unused or changed frequently.

To capture and quantify prescribing practices and dose adjustments, this study evaluated the tolerability of MKIs at VANTHCS. This analysis may also guide clinicians in the selection of starting doses as well as dose titration expectations to optimize MKI therapy.

METHODS

This single-center, retrospective chart review analyzed patients receiving oral oncology MKIs for various malignancies at VANTHCS between January 1, 2014, and October 31, 2024. Participants included adults aged ≥ 18 years with a prescription for axitinib, cabozantinib, lenvatinib, pazopanib, regorafenib, sorafenib, or sunitinib initiated by the hematology/oncology service at VANTHCS. Patients were included if they had follow-up documentation with the hematology/oncology service and/or other VANTHCS clinicians outlining their course of therapy after MKI initiation. Patients were excluded if they did not have sufficient follow-up documentation (eg, transferred care to a non-VA health care practitioner [HCP], moved to another VA health care system), were enrolled in clinical trials, or were prescribed an MKI from a Care in the Community (CITC) prescriber. Electronic health record review and data collection were performed using the VA Computerized Patient Record System and Research Electronic Data Capture. Data were collected from the time of initiation to cessation of therapy and included information regarding therapy changes, progressive disease, and date of death, when available. Data collected included age, sex, race, comorbidities, date of death, type of malignancy and subtypes, cancer stage, MKI used (ie, drug, dose, frequency, schedule, and indication), dates of medication changes (ie, start, adjustment, hold, discontinuation), concurrent antineoplastic treatments, and AEs documented at times of dose change or interruption.

The primary outcome was MKI tolerance determined using relative dose intensity (RDI) and mean and median time on therapy. Two methods are used to calculate RDI that vary in how they approach time on therapy as outlined in Hawn et al.¹¹ This study used method 2, which accounts for holds in therapy by comparing the actual duration of treatment with the duration expected according to treatment protocol. Method 1 compares the prescribed dose with the administered dose and does not adjust for holds.¹¹ Using method 2, the RDI in this study was calculated by dividing the total actual dose given by the total indicated dose for the malignancy being treated, which accounts for duration of treatment.

The total actual dose was the strength, frequency, and days on therapy for each time frame of treatment multiplied together. This method accounted for all dose adjustments and time periods of treatment holds, including patient self-adjustments, prescriber-directed adjustments, and nonadherence determined by HCP documentation and/or prescription data. Similarly, the indicated total dose was calculated by multiplying the indicated strength, frequency, and all days that treatment should have occurred (time from start to finish). Indicated doses were derived from the prescribing information for each malignancy with the exception of sunitinib, for which the off-label dose of 37.5 mg daily was considered a full dose.^12,13 The total indicated dose for axitinib was calculated by considering the dose escalation schedule from the prescribing information.

Patients who required dose reductions due to renal/hepatic impairments or drug-drug interactions had their total indicated dose calculated using dose adjustments listed in the prescribing information. The mean RDI for each MKI agent was calculated by averaging the RDI for each prescription. The overall combined mean RDI included the means of all the MKIs reviewed to avoid skewing the results toward an MKI with more prescriptions. RDIs were also calculated for each cancer type for each agent. Additional descriptive secondary outcomes included rates of AEs and adjustments in doses.

RESULTS

Electronic data extraction identified 278 patients with 366 MKI prescriptions, of which 108 veterans with 158 MKI prescriptions were excluded. The top reason for exclusion was patients managed through CITC. Ultimately, 170 veterans with 208 MKI prescriptions managed by the VANTHCS hematology/oncology clinic were included (Table 1). Among patients receiving MKIs, the mean age was 72.7 years, 98% were male, and 99% had metastatic disease.

The overall combined mean MKI RDI was 67.5% using method 2 and ranged from 85.5% for sunitinib to 49.0% for sorafenib (Figure 1). Additional information regarding mean and median RDIs using method 2 is shown in Figure 1 and further subdivided by cancer type in Table 2. Median RDIs overall were similar to mean RDIs for most agents. Figure 2 indicates the mean and median time on therapy, reflecting time on therapy excluding days therapy was held. The overall combined mean and median days on therapy for all MKIs were 155 days and 95 days, respectively. Mean time on therapy depended on the agent used and ranged from 35 days (regorafenib) to 237 days (cabozantinib).

Of 208 MKI prescriptions, 127 (61.1%) were initiated at a reduced dose due to baseline concerns for tolerance such as performance status, frailty, and prior intolerance of other treatments. Eighty-one prescriptions (38.9%) were initiated at their indicated doses. Ninety prescriptions (43.3%) required dose reductions during treatment. Some MKI prescriptions had multiple dose increases and decreases, which is why RDI more accurately reflects dose adjustments. A total of 376 AEs that contributed to a dose adjustment, hold, or discontinuation occurred across all MKI prescriptions. The most common AEs were 82 failure-to-thrive events (21.8%) (fatigue, malaise, loss of appetite, reduced mobility, global decline), 79 gastrointestinal events (21.0%) (nausea, vomiting, diarrhea, abdominal pain), 62 dermatologic events (16.5%) (rash, hand-foot skin reactions, allergic response), 61 hepatic dysfunction events (16.2%) (liver enzyme elevations, hyperbilirubinemia), 40 cardiovascular events (10.6%) (hypertension, heart failure exacerbations, edema), and 33 renal dysfunction events (8.8%) (acute kidney injury, proteinuria) (Appendix 1).

DISCUSSION

The mean RDI of MKI prescriptions used in the veteran population at VANTHCS was about two-thirds of the indicated dose. These results indicate that most veterans required dose reductions and/or holds due to concerns over initial tolerance/performance status, worsening clinical condition, and/or intolerable AEs attributed to treatment. A retrospective study conducted by Denduluri et al suggested that an RDI of < 85% is a clinically meaningful reduction for traditional chemotherapy based on previous literature.¹⁴ However, it is less clear what RDI should be expected specifically for MKIs in real-world populations. The MKI phase 3 approval trials in RCC for axitinib, lenvatinib, and sunitinib found median RDIs of 89.4%, 69.6% to 70.4%, and 83.9%, respectively. Each trial cited dose reductions most commonly as the result of treatment-related AEs.^15,16

Studies on the impact of RDIs on survival outcomes found that higher RDIs may improve overall and progression-free survival. Retrospective studies inspecting lenvatinib in hepatocellular carcinoma (HCC) indicated that an RDI > 70% in the initial 4 weeks resulted in favorable survival outcomes.¹⁷ Similarly, a retrospective study investigating sunitinib in RCC found that an RDI > 60% conferred favorable survival outcomes.¹⁸ Alghamdi et al noted that patients taking sorafenib for HCC who had RDI > 50% had a favorable trend in survival characteristics. Interestingly, the study found an RDI of 50% to 75% appeared to have better survival than an RDI > 75%.¹⁹ The authors of these studies hypothesized that additional dose reductions allowed for longer total time on therapy due to improved tolerability.^17-19

This analysis found that the RDIs for most MKI agents at VANTHCS were < 85% and lower than the RDIs found in other review articles and phase 3 trials, with the exceptions of pazopanib in thyroid cancer and sunitinib in gastrointestinal stromal tumor (GIST), thyroid cancer, and neuroendocrine cancer. The reasons for the lower RDIs in this study are likely multifactorial, reflecting patient population characteristics, off-label dosing practices, and HCP experiences with these agents. Many veterans have chronic comorbidities that could contribute to reduced performance status and ability to tolerate these therapies. Despite attempts to preemptively reduce doses for patients and account for potential impaired tolerance, there were patients who required further dose reductions in our study.

Failure to thrive was the most common AE leading to dose adjustment or discontinuation, which illustrates the extensive effects these agents have on patient functioning in a real-world population. Notably though, the RDI for sunitinib was higher in this population because about half of patients were dosed using the off-label recommendation, whereas the prescribing information recommends a more intensive 6-week dosing cycle for certain cancer types.^12,13,20 Sorafenib was also often dose-adjusted based on a pharmacokinetic study of sorafenib in renal/hepatic dysfunction, and the RDI likely reflects the off-label prescribing pattern.²¹

Patients with thyroid cancer were found to have higher RDIs compared with those receiving the same agents for other cancer types. Improved tolerability of MKIs in thyroid cancer may be due to a generally more tolerable disease course. Thyroid cancer is the most common cancer in individuals aged < 40 years, a population that is often more robust with fewer comorbidities. Moreover, the 5-year relative survival rate for thyroid cancer remains > 98%.²² This rate is in contrast to those for other cancer types such as HCC, with a 5-year relative survival rate of only 15%.²³

It is challenging to compare the mean and median times on therapy found in this study with those in current literature, as this review included multiple different cancer types for each agent. However, the numbers are generally lower than durations of therapy found across the different disease states and further emphasize the difficulty in tolerating MKIs in the VANTHCS population. Regorafenib had a short duration of time on therapy, which highlights the importance of trials like ReDOS and initiatives such as OCE Project Optimus in helping improve tolerance.^7,8,24

Comparing our results with other studies proved challenging because the RDI calculation methods were not specified. Calculating RDIs in this study using method 1, which does not account for holds, resulted in higher RDIs (Appendix 2). Using method 1, all MKIs had RDIs < 85%, except for pazopanib in thyroid cancer (100%) and RCC (87.9%), and sunitinib in GIST (93.6%), thyroid cancer (100%), and neuroendocrine cancer (93.7%). Notably, using method 1 increased the RDI for pazopanib in neuroendocrine cancer from 5.4% to 50.0%. The low RDI was attributed to a single veteran with a long hold duration, which demonstrates the discrepancy that can occur between the 2 methods.

Limitations

The retrospective design, lack of survival outcomes, and difficulty comparing results with other literature were limitations of this study. Because survival outcomes were not evaluated, future research should seek to investigate how RDIs and dose adjustments made among MKIs can affect survival outcomes in real-world populations. This veteran population with cancer often had multiple chronic comorbidities, which may have contributed to difficulty tolerating MKIs and could have impacted results. Disease-related factors may have influenced the poor tolerance of the MKIs and were not specifically accounted for. Adjustment for comorbidities was not possible because of discrepancies and/or incomplete diagnosis codes and Eastern Cooperative Oncology Group performance status scores documented in patient charts. Therefore, we decided not to report these findings due to potential inaccuracies.

CONCLUSIONS

Results of this study demonstrate that oncology MKI agents used at VANTHCS were difficult for patients to tolerate, leading to suboptimal dosing compared with indicated doses established in clinical trials and prescribing information. Clinicians may use these data to help guide clinical decision-making whenever initiating and managing MKI agents in this population. These findings reinforce that MKI agents are often difficult to tolerate in real-world practice, and indicated doses are often not achieved. Further studies should aim to investigate the effect that various RDIs have on overall survival. Further investigation into different dosing schemes for MKIs to improve tolerability and longer-term use may also prove beneficial.

This analysis may help guide clinicians to carefully approach dosing MKI agents in the veteran population. Given the RDI and AEs, more clinicians may consider starting at lower than indicated doses with the goal to titrate up as tolerated. Additionally, the results highlight the importance of considering palliative care consults and ensuring appropriate supportive care agents are preemptively engaged and adjusted as needed. Approaching dosing and titrations cautiously may help reduce the burden of management on the health care system.

The US Department of Veterans Affairs (VA) annually treats around 450,000 veterans with cancer and diagnoses an additional 56,000.^1,2 Oral multikinase inhibitors (MKIs) are widely used as targeted therapies for many different malignancies. Despite the ease of oral administration, these agents are often accompanied by significant adverse effects (AEs) and drug-drug interactions.^3,4 Common AEs include hypertension, cutaneous reactions, gastrointestinal disturbances, proteinuria, and fatigue. Some serious outcomes that may occur are myocardial infarction, thrombosis, nephrotic syndrome, hemorrhage, hepatotoxicity, and gastrointestinal events.^5,6 Due to poor tolerability of these AEs, dose reductions, frequent therapy holds, and discontinuation of therapy may occur.

The US Food and Drug Administration recognizes dosing challenges with novel therapies and has created the Oncology Center of Excellence (OCE) Project Optimus initiative to reform dose optimization in oncology drug development. The initiative aims to shift the focus from establishing dose regimens based on the maximum tolerated doses of cytotoxic chemotherapeutics to an emphasis on maximum efficacy, safety, and tolerability, which better reflect real-world dosing.^7,8

MKIs can be challenging to manage because of the frequent toxicity-related dose reductions, interruptions, and discontinuations. In a multicenter retrospective study, Schnadig et al investigated dosing characteristics of first-line sunitinib for advanced renal cell carcinoma (RCC) and found that, among 114 patients who experienced AEs while taking sunitinib, 39.5% had dose reductions, 5.3% delayed therapy, 18.4% required additional supportive medications, and 22.8% discontinued sunitinib.⁹ Overall survival and median progression-free survival of these patients were lower than reported by Motzer et al in a phase 3 clinical trial.¹⁰ Schnadig et al concluded that patients treated with sunitinib for RCC in the community setting required more frequent dose reductions and had less time on therapy compared with patients in clinical trials, which ultimately impacted clinical outcomes.⁹

At the VA North Texas Health Care System (VANTHCS), patients with cancer have difficulty tolerating MKIs and often require dose alterations and/or discontinuation because of drug intolerance rather than discontinuation due to progression. Frequent dose adjustments for toxicity management can place more strain on patients and health care resources because of additional appointments, clinician time, and emergency department visits. Escalating drug costs can also cause concern when prescription doses are unused or changed frequently.

To capture and quantify prescribing practices and dose adjustments, this study evaluated the tolerability of MKIs at VANTHCS. This analysis may also guide clinicians in the selection of starting doses as well as dose titration expectations to optimize MKI therapy.

METHODS

This single-center, retrospective chart review analyzed patients receiving oral oncology MKIs for various malignancies at VANTHCS between January 1, 2014, and October 31, 2024. Participants included adults aged ≥ 18 years with a prescription for axitinib, cabozantinib, lenvatinib, pazopanib, regorafenib, sorafenib, or sunitinib initiated by the hematology/oncology service at VANTHCS. Patients were included if they had follow-up documentation with the hematology/oncology service and/or other VANTHCS clinicians outlining their course of therapy after MKI initiation. Patients were excluded if they did not have sufficient follow-up documentation (eg, transferred care to a non-VA health care practitioner [HCP], moved to another VA health care system), were enrolled in clinical trials, or were prescribed an MKI from a Care in the Community (CITC) prescriber. Electronic health record review and data collection were performed using the VA Computerized Patient Record System and Research Electronic Data Capture. Data were collected from the time of initiation to cessation of therapy and included information regarding therapy changes, progressive disease, and date of death, when available. Data collected included age, sex, race, comorbidities, date of death, type of malignancy and subtypes, cancer stage, MKI used (ie, drug, dose, frequency, schedule, and indication), dates of medication changes (ie, start, adjustment, hold, discontinuation), concurrent antineoplastic treatments, and AEs documented at times of dose change or interruption.

The primary outcome was MKI tolerance determined using relative dose intensity (RDI) and mean and median time on therapy. Two methods are used to calculate RDI that vary in how they approach time on therapy as outlined in Hawn et al.¹¹ This study used method 2, which accounts for holds in therapy by comparing the actual duration of treatment with the duration expected according to treatment protocol. Method 1 compares the prescribed dose with the administered dose and does not adjust for holds.¹¹ Using method 2, the RDI in this study was calculated by dividing the total actual dose given by the total indicated dose for the malignancy being treated, which accounts for duration of treatment.

The total actual dose was the strength, frequency, and days on therapy for each time frame of treatment multiplied together. This method accounted for all dose adjustments and time periods of treatment holds, including patient self-adjustments, prescriber-directed adjustments, and nonadherence determined by HCP documentation and/or prescription data. Similarly, the indicated total dose was calculated by multiplying the indicated strength, frequency, and all days that treatment should have occurred (time from start to finish). Indicated doses were derived from the prescribing information for each malignancy with the exception of sunitinib, for which the off-label dose of 37.5 mg daily was considered a full dose.^12,13 The total indicated dose for axitinib was calculated by considering the dose escalation schedule from the prescribing information.

Patients who required dose reductions due to renal/hepatic impairments or drug-drug interactions had their total indicated dose calculated using dose adjustments listed in the prescribing information. The mean RDI for each MKI agent was calculated by averaging the RDI for each prescription. The overall combined mean RDI included the means of all the MKIs reviewed to avoid skewing the results toward an MKI with more prescriptions. RDIs were also calculated for each cancer type for each agent. Additional descriptive secondary outcomes included rates of AEs and adjustments in doses.

RESULTS

Electronic data extraction identified 278 patients with 366 MKI prescriptions, of which 108 veterans with 158 MKI prescriptions were excluded. The top reason for exclusion was patients managed through CITC. Ultimately, 170 veterans with 208 MKI prescriptions managed by the VANTHCS hematology/oncology clinic were included (Table 1). Among patients receiving MKIs, the mean age was 72.7 years, 98% were male, and 99% had metastatic disease.

The overall combined mean MKI RDI was 67.5% using method 2 and ranged from 85.5% for sunitinib to 49.0% for sorafenib (Figure 1). Additional information regarding mean and median RDIs using method 2 is shown in Figure 1 and further subdivided by cancer type in Table 2. Median RDIs overall were similar to mean RDIs for most agents. Figure 2 indicates the mean and median time on therapy, reflecting time on therapy excluding days therapy was held. The overall combined mean and median days on therapy for all MKIs were 155 days and 95 days, respectively. Mean time on therapy depended on the agent used and ranged from 35 days (regorafenib) to 237 days (cabozantinib).

Of 208 MKI prescriptions, 127 (61.1%) were initiated at a reduced dose due to baseline concerns for tolerance such as performance status, frailty, and prior intolerance of other treatments. Eighty-one prescriptions (38.9%) were initiated at their indicated doses. Ninety prescriptions (43.3%) required dose reductions during treatment. Some MKI prescriptions had multiple dose increases and decreases, which is why RDI more accurately reflects dose adjustments. A total of 376 AEs that contributed to a dose adjustment, hold, or discontinuation occurred across all MKI prescriptions. The most common AEs were 82 failure-to-thrive events (21.8%) (fatigue, malaise, loss of appetite, reduced mobility, global decline), 79 gastrointestinal events (21.0%) (nausea, vomiting, diarrhea, abdominal pain), 62 dermatologic events (16.5%) (rash, hand-foot skin reactions, allergic response), 61 hepatic dysfunction events (16.2%) (liver enzyme elevations, hyperbilirubinemia), 40 cardiovascular events (10.6%) (hypertension, heart failure exacerbations, edema), and 33 renal dysfunction events (8.8%) (acute kidney injury, proteinuria) (Appendix 1).

DISCUSSION

The mean RDI of MKI prescriptions used in the veteran population at VANTHCS was about two-thirds of the indicated dose. These results indicate that most veterans required dose reductions and/or holds due to concerns over initial tolerance/performance status, worsening clinical condition, and/or intolerable AEs attributed to treatment. A retrospective study conducted by Denduluri et al suggested that an RDI of < 85% is a clinically meaningful reduction for traditional chemotherapy based on previous literature.¹⁴ However, it is less clear what RDI should be expected specifically for MKIs in real-world populations. The MKI phase 3 approval trials in RCC for axitinib, lenvatinib, and sunitinib found median RDIs of 89.4%, 69.6% to 70.4%, and 83.9%, respectively. Each trial cited dose reductions most commonly as the result of treatment-related AEs.^15,16

Studies on the impact of RDIs on survival outcomes found that higher RDIs may improve overall and progression-free survival. Retrospective studies inspecting lenvatinib in hepatocellular carcinoma (HCC) indicated that an RDI > 70% in the initial 4 weeks resulted in favorable survival outcomes.¹⁷ Similarly, a retrospective study investigating sunitinib in RCC found that an RDI > 60% conferred favorable survival outcomes.¹⁸ Alghamdi et al noted that patients taking sorafenib for HCC who had RDI > 50% had a favorable trend in survival characteristics. Interestingly, the study found an RDI of 50% to 75% appeared to have better survival than an RDI > 75%.¹⁹ The authors of these studies hypothesized that additional dose reductions allowed for longer total time on therapy due to improved tolerability.^17-19

This analysis found that the RDIs for most MKI agents at VANTHCS were < 85% and lower than the RDIs found in other review articles and phase 3 trials, with the exceptions of pazopanib in thyroid cancer and sunitinib in gastrointestinal stromal tumor (GIST), thyroid cancer, and neuroendocrine cancer. The reasons for the lower RDIs in this study are likely multifactorial, reflecting patient population characteristics, off-label dosing practices, and HCP experiences with these agents. Many veterans have chronic comorbidities that could contribute to reduced performance status and ability to tolerate these therapies. Despite attempts to preemptively reduce doses for patients and account for potential impaired tolerance, there were patients who required further dose reductions in our study.

Failure to thrive was the most common AE leading to dose adjustment or discontinuation, which illustrates the extensive effects these agents have on patient functioning in a real-world population. Notably though, the RDI for sunitinib was higher in this population because about half of patients were dosed using the off-label recommendation, whereas the prescribing information recommends a more intensive 6-week dosing cycle for certain cancer types.^12,13,20 Sorafenib was also often dose-adjusted based on a pharmacokinetic study of sorafenib in renal/hepatic dysfunction, and the RDI likely reflects the off-label prescribing pattern.²¹

Patients with thyroid cancer were found to have higher RDIs compared with those receiving the same agents for other cancer types. Improved tolerability of MKIs in thyroid cancer may be due to a generally more tolerable disease course. Thyroid cancer is the most common cancer in individuals aged < 40 years, a population that is often more robust with fewer comorbidities. Moreover, the 5-year relative survival rate for thyroid cancer remains > 98%.²² This rate is in contrast to those for other cancer types such as HCC, with a 5-year relative survival rate of only 15%.²³

It is challenging to compare the mean and median times on therapy found in this study with those in current literature, as this review included multiple different cancer types for each agent. However, the numbers are generally lower than durations of therapy found across the different disease states and further emphasize the difficulty in tolerating MKIs in the VANTHCS population. Regorafenib had a short duration of time on therapy, which highlights the importance of trials like ReDOS and initiatives such as OCE Project Optimus in helping improve tolerance.^7,8,24

Comparing our results with other studies proved challenging because the RDI calculation methods were not specified. Calculating RDIs in this study using method 1, which does not account for holds, resulted in higher RDIs (Appendix 2). Using method 1, all MKIs had RDIs < 85%, except for pazopanib in thyroid cancer (100%) and RCC (87.9%), and sunitinib in GIST (93.6%), thyroid cancer (100%), and neuroendocrine cancer (93.7%). Notably, using method 1 increased the RDI for pazopanib in neuroendocrine cancer from 5.4% to 50.0%. The low RDI was attributed to a single veteran with a long hold duration, which demonstrates the discrepancy that can occur between the 2 methods.

Limitations

The retrospective design, lack of survival outcomes, and difficulty comparing results with other literature were limitations of this study. Because survival outcomes were not evaluated, future research should seek to investigate how RDIs and dose adjustments made among MKIs can affect survival outcomes in real-world populations. This veteran population with cancer often had multiple chronic comorbidities, which may have contributed to difficulty tolerating MKIs and could have impacted results. Disease-related factors may have influenced the poor tolerance of the MKIs and were not specifically accounted for. Adjustment for comorbidities was not possible because of discrepancies and/or incomplete diagnosis codes and Eastern Cooperative Oncology Group performance status scores documented in patient charts. Therefore, we decided not to report these findings due to potential inaccuracies.

CONCLUSIONS

Results of this study demonstrate that oncology MKI agents used at VANTHCS were difficult for patients to tolerate, leading to suboptimal dosing compared with indicated doses established in clinical trials and prescribing information. Clinicians may use these data to help guide clinical decision-making whenever initiating and managing MKI agents in this population. These findings reinforce that MKI agents are often difficult to tolerate in real-world practice, and indicated doses are often not achieved. Further studies should aim to investigate the effect that various RDIs have on overall survival. Further investigation into different dosing schemes for MKIs to improve tolerability and longer-term use may also prove beneficial.

This analysis may help guide clinicians to carefully approach dosing MKI agents in the veteran population. Given the RDI and AEs, more clinicians may consider starting at lower than indicated doses with the goal to titrate up as tolerated. Additionally, the results highlight the importance of considering palliative care consults and ensuring appropriate supportive care agents are preemptively engaged and adjusted as needed. Approaching dosing and titrations cautiously may help reduce the burden of management on the health care system.

Inhaled corticosteroids (ICSs) are frequently prescribed for the treatment of chronic obstructive pulmonary disease (COPD) to reduce exacerbations in a specific subset of patients. The long-term use of ICSs, however, is associated with several potential systemic adverse effects, including adrenal suppression, decreased bone mineral density, and immunosuppression.¹ The concern for immunosuppression is particularly notable and leads to a known increased risk for developing pneumonia in patients with COPD. These patients frequently have other concurrent risk factors for pneumonia (eg, history of tobacco use, older age, and severe airway limitations) and are at higher risk for more severe outcomes in the setting of pneumonia.^2,3

Primarily due to the concern of pneumonia risks, the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines have recommended ICS discontinuation in patients who are less likely to receive significant benefits from therapy.⁴ Likely due to an anti-inflammatory mechanism of action, ICSs have been shown to reduce COPD exacerbation rates in patients with comorbid asthma or who have evidence of a strong inflammatory component to their COPD. The strongest indicator of an inflammatory component is an elevated blood eosinophil (EOS) count; those with EOS > 300 cells/µL are most likely to benefit from ICSs, whereas those with a count < 100 cells/µL are unlikely to have a significant response. In addition to the inflammatory component consideration, prior studies have shown improvements in lung function and reduction of exacerbations with ICS use in patients with frequent moderate-to-severe COPD exacerbations.⁵ Although the GOLD guidelines provide recommendations about who is appropriate to discontinue ICS use, clinicians have no clear guidance on the risks or the best discontinuation strategy.

Based primarily on data from a prior randomized controlled trial, the Veterans Integrated Services Network (VISN) 17, which includes the Veterans Affairs North Texas Health Care System (VANTHCS) in Dallas, established a recommended ICS de-escalation strategy.^6,7 The strategy included a 12-week stepwise taper using a mometasone inhaler for all patients discontinuing a moderate or high dose ICS. The lack of substantial clinical trial data or expert consensus guideline recommendations has left open the question of whether a taper is necessary. To answer that question, this study was conducted to evaluate whether there is a difference in the rate of COPD exacerbations following abrupt discontinuation vs gradual taper of ICS therapy.

Methods

This single-center, retrospective cohort study was conducted at VANTHCS. Patient electronic health records between January 10, 2021, and September 1, 2021, were reviewed for the last documented fill date of any inhaler containing a steroid component. This time frame was chosen to coincide with a VANTHCS initiative to follow GOLD guidelines for ICS discontinuation. Patients were followed for outcomes until November 1, 2022.

To be included in this study, patients had to have active prescriptions at VANTHCS, have a documented diagnosis of COPD in their chart, and be prescribed a stable dose of ICS for ≥ 1 year prior to their latest refill. The inhaler used could contain an ICS as monotherapy, in combination with a long-acting β-agonist (LABA), or as part of triple therapy with an additional long-acting muscarinic antagonist (LAMA). The inhaler needed to be discontinued during the study period of interest.

Patients were excluded if they had a diagnosis of asthma, were aged < 40 years, had active prescriptions for multiple ICS inhalers or nebulizers, or had significant oral steroid use (≥ 5 mg/d prednisone or an equivalent steroid for > 6 weeks) within 1 year of their ICS discontinuation date. In addition, to reduce the risk of future events being misclassified as COPD exacerbations, patients were excluded if they had a congestive heart failure exacerbation up to 2 years before ICS discontinuation or a diagnosis of COVID-19 infection up to 1 year before or 6 months after ICS discontinuation. Patients with a COPD exacerbation requiring an emergency department or hospital visit within 2 years prior to ICS discontinuation were also excluded, as de-escalation of ICS therapy was likely inappropriate in these cases. Finally, patients were excluded if they were started on a different ICS immediately following the discontinuation of their first ICS.

The primary outcome for this study was COPD exacerbations requiring an emergency department visit or hospitalization within 6 months of ICS discontinuation. A secondary outcome examining the rates of COPD exacerbations within 12 months also was used. The original study design called for the use of inferential statistics to compare the rates of primary and secondary outcomes in patients whose ICS was abruptly discontinued with those who were tapered slowly. After data collection, however, the small sample size and low event rate meant that the planned statistical tests were no longer appropriate. Instead, we decided to analyze the planned outcomes using descriptive statistics and look at an additional number of post hoc outcomes to provide deeper insight into clinical practice. We examined the association between relevant demographic factors, such as age, comorbidity burden, ICS potency, duration of ICS therapy, and EOS count and the clinician decision whether to taper the ICS. These same factors were also evaluated for potential association with the increased risk of COPD exacerbations following ICS discontinuation.

 

 

Results

A total of 75 patients were included. Most patients were White race and male with a mean (SD) age of 71.6 (7.4) years. Charlson Comorbidity Index scores were calculated for all included patients with a mean (SD) score of 5.4 (2.0). Of note, scores > 5 are considered a severe comorbidity burden and have an estimated mean 10-year survival rate < 21%. The overwhelming majority of patients were receiving budesonide/formoterol as their ICS inhaler with 1 receiving mometasone monotherapy. When evaluating the steroid dose, 18 (24%) patients received a low dose ICS (200-400 µg of budesonide or 110-220 µg of mometasone), while 57 (76%) received a medium dose (400-800 µg of budesonide or 440 µg of mometasone). No patients received a high ICS dose. The mean (SD) duration of therapy before discontinuation was 4.0 (2.7) years (Table 1).

Nine (12%) patients had their ICS slowly tapered, while therapy was abruptly discontinued in the other 66 (88%) patients. A variety of taper types were used (Figure) without a strong preference for a particular dosing strategy. The primary outcome of COPD exacerbation requiring emergency department visit or hospitalization within 6 months occurred in 2 patients. When the time frame was extended to 12 months for the secondary outcome, an additional 3 patients experienced an event. The mean time to event was 172 days following ICS discontinuation. All the events occurred in patients whose ICS was discontinued without any type of taper.

In a post hoc analysis, we examined the relationship between specific variables and the clinician choice whether to taper an ICS. There was no discernable impact of age, race and ethnicity, comorbidity score, or ICS dose on whether an ICS was tapered. We observed a slight association between shorter duration of therapy and lower EOS count and use of a taper. When evaluating the relationship between these same factors and exacerbation occurrence, we saw comparable trends (Table 2). Patients with an exacerbation had a slightly longer mean duration of ICS therapy and lower mean EOS count.

Discussion

Despite facility guidance recommending tapering of therapy when discontinuing a moderate- or high-dose ICS, most patients in this study discontinued the ICS abruptly. The clinician may have been concerned with patients being able to adhere to a taper regimen, skeptical of the actual need to taper, or unaware of the VANTHCS recommendations for a specific taper method. Shared decision making with patients may have also played a role in prescribing patterns. Currently, there is not sufficient data to support the use of any one particular type of taper over another, which accounts for the variability seen in practice.

The decision to taper ICSs did not seem to be strongly associated with any specific demographic factor, although the ability to examine the impact of factors (eg, race and ethnicity) was limited due to the largely homogenous population. One may have expected a taper to be more common in older patients or in those with more comorbidities; however, this was not observed in this study. The only discernible trends seen were a lower frequency of tapering in patients who had a shorter duration of ICS therapy and those with lower EOS counts. These patients were at lower risk of repeat COPD exacerbations compared with those with longer ICS therapy duration and higher EOS counts; therefore, this finding was unexpected. This suggests that patient-specific factors may not be the primary driving force in the ICS tapering decision; instead it may be based on general clinician preferences or shared decision making with individual patients.

Overall, we noted very low rates of COPD exacerbations. As ICS discontinuation was occurring in stable patients without any recent exacerbations, lower rates of future exacerbations were expected compared with the population of patients with COPD as a whole. This suggests that ICS therapy can be safely stopped in stable patients with COPD who are not likely to receive significant benefits as defined in the GOLD guidelines. All of the exacerbations that occurred were in patients whose ICS was abruptly discontinued; however, given the small number of patients who had a taper, it is difficult to draw conclusions. The low overall rate of exacerbations suggests that a taper may not be necessary to ensure safety while stopping a low- or moderate-intensity ICS.

Several randomized controlled trials have attempted to evaluate the need for an ICS taper; however, results remain mixed. The COSMIC study showed a decline in lung function following ICS discontinuation in patients with ≥ 2 COPD exacerbations in the previous year.⁸ Similar results were seen in the SUNSET study with increased exacerbation rates after ICS discontinuation in patients with elevated EOS counts.⁹ However, these studies included patients for whom ICS discontinuation is currently not recommended. Alternatively, the INSTEAD trial looked at patients without frequent recent exacerbations and found no difference in lung function, exacerbation rates, or rescue inhaler use in patients that continued combination ICS plus bronchodilator use vs those de-escalated to bronchodilator monotherapy.¹⁰

All 3 studies chose to abruptly stop the ICS when discontinuing therapy; however, using a slow, stepwise taper similar to that used after long periods of oral steroid use may reduce the risk of worsening exacerbations. The WISDOM trial is the only major randomized trial to date that stopped ICS therapy using a stepwise withdrawal of therapy.⁷ In patients who were continued on triple inhaled therapy (2 bronchodilators plus ICS) vs those who were de-escalated to dual bronchodilator therapy, de-escalation was noninferior to continuation of therapy in time to first COPD exacerbation. Both the WISDOM and INSTEAD trials were consistent with the results found in our real-world retrospective evaluation.

There did not seem to be an increased exacerbation risk following ICS discontinuation in any patient subpopulation based on sex, age, race and ethnicity, or comorbidity burden. We noted a trend toward more exacerbations in patients with a longer duration of ICS therapy, suggesting that additional caution may be needed when stopping ICS therapy for these patients. We also noted a trend toward more exacerbations in patients with a lower mean EOS count; however, given the low event rate and wide variability in observed patient EOS counts, this is likely a spurious finding.

 

Limitations

The small sample size, resulting from the strict exclusion criteria, limits the generalizability of the results. Although the low number of events seen in this study supports safety in ICS discontinuation, there may have been higher rates observed in a larger population. The most common reason for patient exclusion was the initiation of another ICS immediately following discontinuation of the original ICS. During the study period, VANTHCS underwent a change to its formulary: Fluticasone/salmeterol replaced budesonide/formoterol as the preferred ICS/LABA combination. As a result, many patients had their budesonide/formoterol discontinued during the study period solely to initiate fluticasone/salmeterol therapy. As these patients did not truly have their ICS discontinued or have a significant period without ICS therapy, they were not included in the results, and the total patient population available to analyze was relatively limited.

The low event rate also limits the ability to compare various factors influencing exacerbation risk, particularly taper vs abrupt ICS discontinuation. This is further compounded by the small number of patients who had a taper performed and the lack of consistency in the method of tapering used. Statistical significance could not be determined for any outcome, and all findings were purely hypothesis generating. Finally, data were only collected for moderate or severe COPD exacerbations that resulted in an emergency department visit or hospitalization, so there may have been mild exacerbations treated in the outpatient setting that were not captured.

Despite these limitations, this study adds data to an area of COPD management that currently lacks strong clinical guidance. Since investigators had access to clinician notes, we were able to capture ICS tapers even if patients did not receive a prescription with specific taper instructions. The extended follow-up period of 12 months evaluated a longer potential time to impact of ICS discontinuation than is done in most COPD clinical trials.

Conclusions

Overall, very low rates of COPD exacerbations occurred following ICS discontinuation, regardless of whether a taper was used. The results suggest that there may be several appropriate ways to discontinue ICS therapy. However, there is insufficient evidence to support a particular taper or the need to taper at all. It seems to be safe to discontinue ICS therapy in patients who are unlikely to benefit from continued use; however, patient-specific factors should be considered as part of clinical decision making.

Inhaled corticosteroids (ICSs) are frequently prescribed for the treatment of chronic obstructive pulmonary disease (COPD) to reduce exacerbations in a specific subset of patients. The long-term use of ICSs, however, is associated with several potential systemic adverse effects, including adrenal suppression, decreased bone mineral density, and immunosuppression.¹ The concern for immunosuppression is particularly notable and leads to a known increased risk for developing pneumonia in patients with COPD. These patients frequently have other concurrent risk factors for pneumonia (eg, history of tobacco use, older age, and severe airway limitations) and are at higher risk for more severe outcomes in the setting of pneumonia.^2,3

Primarily due to the concern of pneumonia risks, the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines have recommended ICS discontinuation in patients who are less likely to receive significant benefits from therapy.⁴ Likely due to an anti-inflammatory mechanism of action, ICSs have been shown to reduce COPD exacerbation rates in patients with comorbid asthma or who have evidence of a strong inflammatory component to their COPD. The strongest indicator of an inflammatory component is an elevated blood eosinophil (EOS) count; those with EOS > 300 cells/µL are most likely to benefit from ICSs, whereas those with a count < 100 cells/µL are unlikely to have a significant response. In addition to the inflammatory component consideration, prior studies have shown improvements in lung function and reduction of exacerbations with ICS use in patients with frequent moderate-to-severe COPD exacerbations.⁵ Although the GOLD guidelines provide recommendations about who is appropriate to discontinue ICS use, clinicians have no clear guidance on the risks or the best discontinuation strategy.

Based primarily on data from a prior randomized controlled trial, the Veterans Integrated Services Network (VISN) 17, which includes the Veterans Affairs North Texas Health Care System (VANTHCS) in Dallas, established a recommended ICS de-escalation strategy.^6,7 The strategy included a 12-week stepwise taper using a mometasone inhaler for all patients discontinuing a moderate or high dose ICS. The lack of substantial clinical trial data or expert consensus guideline recommendations has left open the question of whether a taper is necessary. To answer that question, this study was conducted to evaluate whether there is a difference in the rate of COPD exacerbations following abrupt discontinuation vs gradual taper of ICS therapy.

Methods

This single-center, retrospective cohort study was conducted at VANTHCS. Patient electronic health records between January 10, 2021, and September 1, 2021, were reviewed for the last documented fill date of any inhaler containing a steroid component. This time frame was chosen to coincide with a VANTHCS initiative to follow GOLD guidelines for ICS discontinuation. Patients were followed for outcomes until November 1, 2022.

To be included in this study, patients had to have active prescriptions at VANTHCS, have a documented diagnosis of COPD in their chart, and be prescribed a stable dose of ICS for ≥ 1 year prior to their latest refill. The inhaler used could contain an ICS as monotherapy, in combination with a long-acting β-agonist (LABA), or as part of triple therapy with an additional long-acting muscarinic antagonist (LAMA). The inhaler needed to be discontinued during the study period of interest.

Patients were excluded if they had a diagnosis of asthma, were aged < 40 years, had active prescriptions for multiple ICS inhalers or nebulizers, or had significant oral steroid use (≥ 5 mg/d prednisone or an equivalent steroid for > 6 weeks) within 1 year of their ICS discontinuation date. In addition, to reduce the risk of future events being misclassified as COPD exacerbations, patients were excluded if they had a congestive heart failure exacerbation up to 2 years before ICS discontinuation or a diagnosis of COVID-19 infection up to 1 year before or 6 months after ICS discontinuation. Patients with a COPD exacerbation requiring an emergency department or hospital visit within 2 years prior to ICS discontinuation were also excluded, as de-escalation of ICS therapy was likely inappropriate in these cases. Finally, patients were excluded if they were started on a different ICS immediately following the discontinuation of their first ICS.

The primary outcome for this study was COPD exacerbations requiring an emergency department visit or hospitalization within 6 months of ICS discontinuation. A secondary outcome examining the rates of COPD exacerbations within 12 months also was used. The original study design called for the use of inferential statistics to compare the rates of primary and secondary outcomes in patients whose ICS was abruptly discontinued with those who were tapered slowly. After data collection, however, the small sample size and low event rate meant that the planned statistical tests were no longer appropriate. Instead, we decided to analyze the planned outcomes using descriptive statistics and look at an additional number of post hoc outcomes to provide deeper insight into clinical practice. We examined the association between relevant demographic factors, such as age, comorbidity burden, ICS potency, duration of ICS therapy, and EOS count and the clinician decision whether to taper the ICS. These same factors were also evaluated for potential association with the increased risk of COPD exacerbations following ICS discontinuation.

 

 

Results

A total of 75 patients were included. Most patients were White race and male with a mean (SD) age of 71.6 (7.4) years. Charlson Comorbidity Index scores were calculated for all included patients with a mean (SD) score of 5.4 (2.0). Of note, scores > 5 are considered a severe comorbidity burden and have an estimated mean 10-year survival rate < 21%. The overwhelming majority of patients were receiving budesonide/formoterol as their ICS inhaler with 1 receiving mometasone monotherapy. When evaluating the steroid dose, 18 (24%) patients received a low dose ICS (200-400 µg of budesonide or 110-220 µg of mometasone), while 57 (76%) received a medium dose (400-800 µg of budesonide or 440 µg of mometasone). No patients received a high ICS dose. The mean (SD) duration of therapy before discontinuation was 4.0 (2.7) years (Table 1).

Nine (12%) patients had their ICS slowly tapered, while therapy was abruptly discontinued in the other 66 (88%) patients. A variety of taper types were used (Figure) without a strong preference for a particular dosing strategy. The primary outcome of COPD exacerbation requiring emergency department visit or hospitalization within 6 months occurred in 2 patients. When the time frame was extended to 12 months for the secondary outcome, an additional 3 patients experienced an event. The mean time to event was 172 days following ICS discontinuation. All the events occurred in patients whose ICS was discontinued without any type of taper.

In a post hoc analysis, we examined the relationship between specific variables and the clinician choice whether to taper an ICS. There was no discernable impact of age, race and ethnicity, comorbidity score, or ICS dose on whether an ICS was tapered. We observed a slight association between shorter duration of therapy and lower EOS count and use of a taper. When evaluating the relationship between these same factors and exacerbation occurrence, we saw comparable trends (Table 2). Patients with an exacerbation had a slightly longer mean duration of ICS therapy and lower mean EOS count.

Discussion

Despite facility guidance recommending tapering of therapy when discontinuing a moderate- or high-dose ICS, most patients in this study discontinued the ICS abruptly. The clinician may have been concerned with patients being able to adhere to a taper regimen, skeptical of the actual need to taper, or unaware of the VANTHCS recommendations for a specific taper method. Shared decision making with patients may have also played a role in prescribing patterns. Currently, there is not sufficient data to support the use of any one particular type of taper over another, which accounts for the variability seen in practice.

The decision to taper ICSs did not seem to be strongly associated with any specific demographic factor, although the ability to examine the impact of factors (eg, race and ethnicity) was limited due to the largely homogenous population. One may have expected a taper to be more common in older patients or in those with more comorbidities; however, this was not observed in this study. The only discernible trends seen were a lower frequency of tapering in patients who had a shorter duration of ICS therapy and those with lower EOS counts. These patients were at lower risk of repeat COPD exacerbations compared with those with longer ICS therapy duration and higher EOS counts; therefore, this finding was unexpected. This suggests that patient-specific factors may not be the primary driving force in the ICS tapering decision; instead it may be based on general clinician preferences or shared decision making with individual patients.

Overall, we noted very low rates of COPD exacerbations. As ICS discontinuation was occurring in stable patients without any recent exacerbations, lower rates of future exacerbations were expected compared with the population of patients with COPD as a whole. This suggests that ICS therapy can be safely stopped in stable patients with COPD who are not likely to receive significant benefits as defined in the GOLD guidelines. All of the exacerbations that occurred were in patients whose ICS was abruptly discontinued; however, given the small number of patients who had a taper, it is difficult to draw conclusions. The low overall rate of exacerbations suggests that a taper may not be necessary to ensure safety while stopping a low- or moderate-intensity ICS.

Several randomized controlled trials have attempted to evaluate the need for an ICS taper; however, results remain mixed. The COSMIC study showed a decline in lung function following ICS discontinuation in patients with ≥ 2 COPD exacerbations in the previous year.⁸ Similar results were seen in the SUNSET study with increased exacerbation rates after ICS discontinuation in patients with elevated EOS counts.⁹ However, these studies included patients for whom ICS discontinuation is currently not recommended. Alternatively, the INSTEAD trial looked at patients without frequent recent exacerbations and found no difference in lung function, exacerbation rates, or rescue inhaler use in patients that continued combination ICS plus bronchodilator use vs those de-escalated to bronchodilator monotherapy.¹⁰

All 3 studies chose to abruptly stop the ICS when discontinuing therapy; however, using a slow, stepwise taper similar to that used after long periods of oral steroid use may reduce the risk of worsening exacerbations. The WISDOM trial is the only major randomized trial to date that stopped ICS therapy using a stepwise withdrawal of therapy.⁷ In patients who were continued on triple inhaled therapy (2 bronchodilators plus ICS) vs those who were de-escalated to dual bronchodilator therapy, de-escalation was noninferior to continuation of therapy in time to first COPD exacerbation. Both the WISDOM and INSTEAD trials were consistent with the results found in our real-world retrospective evaluation.

There did not seem to be an increased exacerbation risk following ICS discontinuation in any patient subpopulation based on sex, age, race and ethnicity, or comorbidity burden. We noted a trend toward more exacerbations in patients with a longer duration of ICS therapy, suggesting that additional caution may be needed when stopping ICS therapy for these patients. We also noted a trend toward more exacerbations in patients with a lower mean EOS count; however, given the low event rate and wide variability in observed patient EOS counts, this is likely a spurious finding.

 

Limitations

The small sample size, resulting from the strict exclusion criteria, limits the generalizability of the results. Although the low number of events seen in this study supports safety in ICS discontinuation, there may have been higher rates observed in a larger population. The most common reason for patient exclusion was the initiation of another ICS immediately following discontinuation of the original ICS. During the study period, VANTHCS underwent a change to its formulary: Fluticasone/salmeterol replaced budesonide/formoterol as the preferred ICS/LABA combination. As a result, many patients had their budesonide/formoterol discontinued during the study period solely to initiate fluticasone/salmeterol therapy. As these patients did not truly have their ICS discontinued or have a significant period without ICS therapy, they were not included in the results, and the total patient population available to analyze was relatively limited.

The low event rate also limits the ability to compare various factors influencing exacerbation risk, particularly taper vs abrupt ICS discontinuation. This is further compounded by the small number of patients who had a taper performed and the lack of consistency in the method of tapering used. Statistical significance could not be determined for any outcome, and all findings were purely hypothesis generating. Finally, data were only collected for moderate or severe COPD exacerbations that resulted in an emergency department visit or hospitalization, so there may have been mild exacerbations treated in the outpatient setting that were not captured.

Despite these limitations, this study adds data to an area of COPD management that currently lacks strong clinical guidance. Since investigators had access to clinician notes, we were able to capture ICS tapers even if patients did not receive a prescription with specific taper instructions. The extended follow-up period of 12 months evaluated a longer potential time to impact of ICS discontinuation than is done in most COPD clinical trials.

Conclusions

Overall, very low rates of COPD exacerbations occurred following ICS discontinuation, regardless of whether a taper was used. The results suggest that there may be several appropriate ways to discontinue ICS therapy. However, there is insufficient evidence to support a particular taper or the need to taper at all. It seems to be safe to discontinue ICS therapy in patients who are unlikely to benefit from continued use; however, patient-specific factors should be considered as part of clinical decision making.

Inhaled corticosteroids (ICSs) are frequently prescribed for the treatment of chronic obstructive pulmonary disease (COPD) to reduce exacerbations in a specific subset of patients. The long-term use of ICSs, however, is associated with several potential systemic adverse effects, including adrenal suppression, decreased bone mineral density, and immunosuppression.¹ The concern for immunosuppression is particularly notable and leads to a known increased risk for developing pneumonia in patients with COPD. These patients frequently have other concurrent risk factors for pneumonia (eg, history of tobacco use, older age, and severe airway limitations) and are at higher risk for more severe outcomes in the setting of pneumonia.^2,3

Primarily due to the concern of pneumonia risks, the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines have recommended ICS discontinuation in patients who are less likely to receive significant benefits from therapy.⁴ Likely due to an anti-inflammatory mechanism of action, ICSs have been shown to reduce COPD exacerbation rates in patients with comorbid asthma or who have evidence of a strong inflammatory component to their COPD. The strongest indicator of an inflammatory component is an elevated blood eosinophil (EOS) count; those with EOS > 300 cells/µL are most likely to benefit from ICSs, whereas those with a count < 100 cells/µL are unlikely to have a significant response. In addition to the inflammatory component consideration, prior studies have shown improvements in lung function and reduction of exacerbations with ICS use in patients with frequent moderate-to-severe COPD exacerbations.⁵ Although the GOLD guidelines provide recommendations about who is appropriate to discontinue ICS use, clinicians have no clear guidance on the risks or the best discontinuation strategy.

Based primarily on data from a prior randomized controlled trial, the Veterans Integrated Services Network (VISN) 17, which includes the Veterans Affairs North Texas Health Care System (VANTHCS) in Dallas, established a recommended ICS de-escalation strategy.^6,7 The strategy included a 12-week stepwise taper using a mometasone inhaler for all patients discontinuing a moderate or high dose ICS. The lack of substantial clinical trial data or expert consensus guideline recommendations has left open the question of whether a taper is necessary. To answer that question, this study was conducted to evaluate whether there is a difference in the rate of COPD exacerbations following abrupt discontinuation vs gradual taper of ICS therapy.

Methods

This single-center, retrospective cohort study was conducted at VANTHCS. Patient electronic health records between January 10, 2021, and September 1, 2021, were reviewed for the last documented fill date of any inhaler containing a steroid component. This time frame was chosen to coincide with a VANTHCS initiative to follow GOLD guidelines for ICS discontinuation. Patients were followed for outcomes until November 1, 2022.

To be included in this study, patients had to have active prescriptions at VANTHCS, have a documented diagnosis of COPD in their chart, and be prescribed a stable dose of ICS for ≥ 1 year prior to their latest refill. The inhaler used could contain an ICS as monotherapy, in combination with a long-acting β-agonist (LABA), or as part of triple therapy with an additional long-acting muscarinic antagonist (LAMA). The inhaler needed to be discontinued during the study period of interest.

Patients were excluded if they had a diagnosis of asthma, were aged < 40 years, had active prescriptions for multiple ICS inhalers or nebulizers, or had significant oral steroid use (≥ 5 mg/d prednisone or an equivalent steroid for > 6 weeks) within 1 year of their ICS discontinuation date. In addition, to reduce the risk of future events being misclassified as COPD exacerbations, patients were excluded if they had a congestive heart failure exacerbation up to 2 years before ICS discontinuation or a diagnosis of COVID-19 infection up to 1 year before or 6 months after ICS discontinuation. Patients with a COPD exacerbation requiring an emergency department or hospital visit within 2 years prior to ICS discontinuation were also excluded, as de-escalation of ICS therapy was likely inappropriate in these cases. Finally, patients were excluded if they were started on a different ICS immediately following the discontinuation of their first ICS.

The primary outcome for this study was COPD exacerbations requiring an emergency department visit or hospitalization within 6 months of ICS discontinuation. A secondary outcome examining the rates of COPD exacerbations within 12 months also was used. The original study design called for the use of inferential statistics to compare the rates of primary and secondary outcomes in patients whose ICS was abruptly discontinued with those who were tapered slowly. After data collection, however, the small sample size and low event rate meant that the planned statistical tests were no longer appropriate. Instead, we decided to analyze the planned outcomes using descriptive statistics and look at an additional number of post hoc outcomes to provide deeper insight into clinical practice. We examined the association between relevant demographic factors, such as age, comorbidity burden, ICS potency, duration of ICS therapy, and EOS count and the clinician decision whether to taper the ICS. These same factors were also evaluated for potential association with the increased risk of COPD exacerbations following ICS discontinuation.

 

 

Results

A total of 75 patients were included. Most patients were White race and male with a mean (SD) age of 71.6 (7.4) years. Charlson Comorbidity Index scores were calculated for all included patients with a mean (SD) score of 5.4 (2.0). Of note, scores > 5 are considered a severe comorbidity burden and have an estimated mean 10-year survival rate < 21%. The overwhelming majority of patients were receiving budesonide/formoterol as their ICS inhaler with 1 receiving mometasone monotherapy. When evaluating the steroid dose, 18 (24%) patients received a low dose ICS (200-400 µg of budesonide or 110-220 µg of mometasone), while 57 (76%) received a medium dose (400-800 µg of budesonide or 440 µg of mometasone). No patients received a high ICS dose. The mean (SD) duration of therapy before discontinuation was 4.0 (2.7) years (Table 1).

Nine (12%) patients had their ICS slowly tapered, while therapy was abruptly discontinued in the other 66 (88%) patients. A variety of taper types were used (Figure) without a strong preference for a particular dosing strategy. The primary outcome of COPD exacerbation requiring emergency department visit or hospitalization within 6 months occurred in 2 patients. When the time frame was extended to 12 months for the secondary outcome, an additional 3 patients experienced an event. The mean time to event was 172 days following ICS discontinuation. All the events occurred in patients whose ICS was discontinued without any type of taper.

In a post hoc analysis, we examined the relationship between specific variables and the clinician choice whether to taper an ICS. There was no discernable impact of age, race and ethnicity, comorbidity score, or ICS dose on whether an ICS was tapered. We observed a slight association between shorter duration of therapy and lower EOS count and use of a taper. When evaluating the relationship between these same factors and exacerbation occurrence, we saw comparable trends (Table 2). Patients with an exacerbation had a slightly longer mean duration of ICS therapy and lower mean EOS count.

Discussion

Despite facility guidance recommending tapering of therapy when discontinuing a moderate- or high-dose ICS, most patients in this study discontinued the ICS abruptly. The clinician may have been concerned with patients being able to adhere to a taper regimen, skeptical of the actual need to taper, or unaware of the VANTHCS recommendations for a specific taper method. Shared decision making with patients may have also played a role in prescribing patterns. Currently, there is not sufficient data to support the use of any one particular type of taper over another, which accounts for the variability seen in practice.

The decision to taper ICSs did not seem to be strongly associated with any specific demographic factor, although the ability to examine the impact of factors (eg, race and ethnicity) was limited due to the largely homogenous population. One may have expected a taper to be more common in older patients or in those with more comorbidities; however, this was not observed in this study. The only discernible trends seen were a lower frequency of tapering in patients who had a shorter duration of ICS therapy and those with lower EOS counts. These patients were at lower risk of repeat COPD exacerbations compared with those with longer ICS therapy duration and higher EOS counts; therefore, this finding was unexpected. This suggests that patient-specific factors may not be the primary driving force in the ICS tapering decision; instead it may be based on general clinician preferences or shared decision making with individual patients.

Overall, we noted very low rates of COPD exacerbations. As ICS discontinuation was occurring in stable patients without any recent exacerbations, lower rates of future exacerbations were expected compared with the population of patients with COPD as a whole. This suggests that ICS therapy can be safely stopped in stable patients with COPD who are not likely to receive significant benefits as defined in the GOLD guidelines. All of the exacerbations that occurred were in patients whose ICS was abruptly discontinued; however, given the small number of patients who had a taper, it is difficult to draw conclusions. The low overall rate of exacerbations suggests that a taper may not be necessary to ensure safety while stopping a low- or moderate-intensity ICS.

Several randomized controlled trials have attempted to evaluate the need for an ICS taper; however, results remain mixed. The COSMIC study showed a decline in lung function following ICS discontinuation in patients with ≥ 2 COPD exacerbations in the previous year.⁸ Similar results were seen in the SUNSET study with increased exacerbation rates after ICS discontinuation in patients with elevated EOS counts.⁹ However, these studies included patients for whom ICS discontinuation is currently not recommended. Alternatively, the INSTEAD trial looked at patients without frequent recent exacerbations and found no difference in lung function, exacerbation rates, or rescue inhaler use in patients that continued combination ICS plus bronchodilator use vs those de-escalated to bronchodilator monotherapy.¹⁰

All 3 studies chose to abruptly stop the ICS when discontinuing therapy; however, using a slow, stepwise taper similar to that used after long periods of oral steroid use may reduce the risk of worsening exacerbations. The WISDOM trial is the only major randomized trial to date that stopped ICS therapy using a stepwise withdrawal of therapy.⁷ In patients who were continued on triple inhaled therapy (2 bronchodilators plus ICS) vs those who were de-escalated to dual bronchodilator therapy, de-escalation was noninferior to continuation of therapy in time to first COPD exacerbation. Both the WISDOM and INSTEAD trials were consistent with the results found in our real-world retrospective evaluation.

There did not seem to be an increased exacerbation risk following ICS discontinuation in any patient subpopulation based on sex, age, race and ethnicity, or comorbidity burden. We noted a trend toward more exacerbations in patients with a longer duration of ICS therapy, suggesting that additional caution may be needed when stopping ICS therapy for these patients. We also noted a trend toward more exacerbations in patients with a lower mean EOS count; however, given the low event rate and wide variability in observed patient EOS counts, this is likely a spurious finding.

 

Limitations

The small sample size, resulting from the strict exclusion criteria, limits the generalizability of the results. Although the low number of events seen in this study supports safety in ICS discontinuation, there may have been higher rates observed in a larger population. The most common reason for patient exclusion was the initiation of another ICS immediately following discontinuation of the original ICS. During the study period, VANTHCS underwent a change to its formulary: Fluticasone/salmeterol replaced budesonide/formoterol as the preferred ICS/LABA combination. As a result, many patients had their budesonide/formoterol discontinued during the study period solely to initiate fluticasone/salmeterol therapy. As these patients did not truly have their ICS discontinued or have a significant period without ICS therapy, they were not included in the results, and the total patient population available to analyze was relatively limited.

The low event rate also limits the ability to compare various factors influencing exacerbation risk, particularly taper vs abrupt ICS discontinuation. This is further compounded by the small number of patients who had a taper performed and the lack of consistency in the method of tapering used. Statistical significance could not be determined for any outcome, and all findings were purely hypothesis generating. Finally, data were only collected for moderate or severe COPD exacerbations that resulted in an emergency department visit or hospitalization, so there may have been mild exacerbations treated in the outpatient setting that were not captured.

Despite these limitations, this study adds data to an area of COPD management that currently lacks strong clinical guidance. Since investigators had access to clinician notes, we were able to capture ICS tapers even if patients did not receive a prescription with specific taper instructions. The extended follow-up period of 12 months evaluated a longer potential time to impact of ICS discontinuation than is done in most COPD clinical trials.

Conclusions

Overall, very low rates of COPD exacerbations occurred following ICS discontinuation, regardless of whether a taper was used. The results suggest that there may be several appropriate ways to discontinue ICS therapy. However, there is insufficient evidence to support a particular taper or the need to taper at all. It seems to be safe to discontinue ICS therapy in patients who are unlikely to benefit from continued use; however, patient-specific factors should be considered as part of clinical decision making.