Immune-Related Adverse Events Requiring Hospitalization: Spectrum of Toxicity, Treatment, and Outcomes

Abstract

PURPOSE:

Immune checkpoint inhibitors (ICIs) cause immune-related adverse events (irAEs). The proportion of patients who are hospitalized for irAEs and their spectrum, management, and outcomes are not well described.

METHODS:

We report the proportion of hospitalized patients in an academic center who were treated with ICIs from May to December 2017. Patient characteristics, toxicities, management, and outcomes for confirmed irAE admissions are reported. Associations between patient features and irAE hospitalizations are examined.

RESULTS:

Twenty-three percent (n = 100) of 443 patients who were admitted to an academic oncology center over 6 months had ever received ICIs. Of these patients, 41% were admitted for suspected irAEs and 23% were confirmed irAEs. IrAEs accounted for 5% of all oncology hospitalizations (n = 23). Ninety-one percent of patients with confirmed irAEs prompted a medicine subspecialist consultation, most commonly gastroenterology (22%). Fifteen patients (65%) had their irAEs improve/resolve, seven (30%) had worsening irAEs, and three (13%) died of their irAEs. The majority of patients (n = 20; 87%) discontinued ICIs after discharge. Among ICI-treated patients who required admission, an increased likelihood of irAE-related hospitalization was associated with patient age older than 65 years (odds ratio, 5.4; 95% CI, 1.6 to 17.8) and receipt of combination immunotherapy (OR, 6.8; 95% CI, 2.0 to 23.2).

CONCLUSION:

A notable proportion of ICI-treated patients are hospitalized for irAEs, and these patients have a high demand for multidisciplinary management. Older age and combination ICI treatment were associated with an increased risk of irAE-related hospitalization. Whereas these data are from an academic center and include patients in clinical trials, with expanding use of ICIs, these data have important implications for inpatient service planning and risk stratification.

INTRODUCTION

Immune checkpoint pathways are a system of checks and balances that modulate immunologic self-activity.¹ Tumor cells exploit these pathways to evade detection and destruction by the host immune system.¹ Immune checkpoint inhibitors (ICIs) are anticancer agents that modulate these pathways to reinvigorate an anticancer immune response.² In particular, ICIs against programmed death–1 (PD-1),³ programmed death-ligand 1 (PD-L1),⁴ and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4)³ have demonstrated efficacy in multiple cancers.⁵ However, ICI therapy is associated with the development of immune-related adverse events (irAEs).⁶ Our current understanding of the spectrum and severity of irAEs is largely on the basis of clinical trial data and meta-analyses.^6-9 The most common irAEs from CTLA-4 inhibitors include diarrhea/colitis and dermatitis,⁹ whereas PD-1/PD-L1 inhibitors are associated with dermatitis, thyroid dysfunction, and pneumonitis.¹⁰ The incidence of irAEs is higher in patients who are treated with anti–CTLA-4 plus anti–PD-1/PD-L1 combinations, rather than monotherapy with either agent alone.¹¹ The severity of irAEs is graded by the Common Terminology Criteria for Adverse Events, version 4.03,¹² such that grade 1 toxicity is treated supportively either by ICI withholding or continued dosing with close monitoring,¹³ grade 2 irAEs with corticosteroids and ICI withholding, and grade 3 irAEs or higher with hospitalization, intravenous corticosteroids, and occasionally additional immunosuppression.^14-16 Oncologic societies have published guidelines to direct organ-specific irAE diagnosis and management.^17-19

To date, the clinical features, range of irAEs by organ system, irAE severity, hospital-based services needed to support these patients, and outcomes of patients who have been hospitalized for irAEs have not been well described. We aimed to identify the proportion of hospitalized patients in an academic oncology center who had received ICIs and characterize the spectrum of toxicities, management, and outcomes of hospitalizations for irAEs. In addition, we describe risk factors for hospitalization and patterns of subspecialist referral to provide insight into the service needs for this growing patient population.

METHODS

Patients

Patients with solid tumor malignancies who were hospitalized on the oncology service at an academic center (Johns Hopkins Hospital [JHH]) between May 31, 2017, and December 31, 2017, were identified in real time by one abstractor (A.B.) and confirmed by a second abstractor (J.N.). Those who had ever received ICIs, either as standard of care or in clinical trials, were identified. Patients were excluded if they were not admitted at JHH and were excluded from the ICI-treated group if they had received non-ICI immunotherapy alone, such as cancer vaccines.

Data Collection and Variable Definition

A confirmed irAE was defined as a toxicity in which there was consensus on irAE diagnosis between the admitting and primary oncologists, clinical improvement with irAE-based management, exclusion of alternative diagnoses, and/or pathologic evidence of irAE, where available. Chronic irAEs were defined as those that required medical management for more than 6 months, including lifelong hormonal replacement.²⁰ Admission referred to the first hospitalization in the study period, whereas readmission referred to the second admission within the same study period.

Statistical Analysis

Patient, tumor, oncologic therapy, and irAE data were collected in an institutional review board–approved database and summarized using descriptive statistics. Categorical and continuous variables were compared using χ² and Wilcoxon rank sum tests, respectively. Multivariable logistic regression was used to obtain odds ratios (ORs) for the independent effects of age (< 65 v ≥ 65 years old), sex (male v female), smoking status (never v ever/current smoker), treatment (monotherapy v combination therapy [anti–PD-1/anti–PD-L1 and anti–CTLA-4]), history of irAE (yes v no), prior autoimmune disease (yes v no), and clinical trial status (treatment received in clinical trial v standard of care) on irAE-related hospitalizations among ICI-treated patients who required admission. Statistical analyses were performed using STATA 14.1 software (STATA, College Station, TX; Computing Resource Center, Santa Monica, CA). Statistical tests were two sided with a P value of < .05 indicating statistical significance.

RESULTS

Patient Cohort

Over a 6-month period, 443 individual patients were admitted to JHH oncology (Fig A1, online only). One hundred patients (23%) had ever received an ICI. Of these patients, 26 were actively receiving ICIs at the time of admission (26%), whereas 74 had either completed or stopped treatment (74%). The ICI regimen most commonly used was anti–PD-1/PD-L1 monotherapy—received by 69% of patients—and the agents most commonly used were nivolumab (34%) and pembrolizumab (23%). Patients received a median of one ICI dose before hospitalization (range, 1 to 73; ≤ 10 doses: n = 79; 11 to 20 doses: n = 13; 21 to 30 doses: n = 5; 31 to 40 doses: n = 2; ≥ 40 doses: n = 1). Median time from beginning ICIs to irAE hospitalization was 64 days (range, 13 to 566 days) and the most remote ICI treatment was administered 2.5 years before patient admission.

Fig 1.

Associations between clinical features and immune-related adverse event (irAE) admission. Several patient factors were identified and assessed as to whether their presence increased the likelihood of inpatient admission for an irAE. ICI, immune checkpoint inhibitor; OR, odds ratio.

Twenty-five patients (25%) had known autoimmune condition(s) before initiating ICIs, including hypothyroidism (56%), thyroiditis (4%), inflammatory bowel disease (8%), inflammatory arthritis (16%), psoriasis (8%), focal sclerosing glomerulonephritis (4%), and systemic lupus erythematosus (4%). Of 100 ICI-treated patients, 41 (41%), were hospitalized for a suspected irAE and 23 (23%) had confirmed irAEs. Separately, 16% of patients with prior autoimmune conditions experienced a flare of their condition and two (8%) warranted hospitalization. These events were not deemed irAEs.

Among the 59 ICI-treated patients who were admitted without suspicion of irAE, an irAE was subsequently confirmed as the reason for hospitalization in three patients (5%). Therefore, the cumulative incidence of confirmed irAEs among ICI-treated patients who were admitted to oncology was 23% over 6 months. Other reasons for hospitalization among ICI-treated patients were complications of primary cancer (52%), systemic chemotherapy effects (2%), infection (21%), surgery (1%), or elective cancer therapy (1%; Fig A1).

irAE Admissions

Patient, tumor, and treatment characteristics for ICI-treated patients, stratified by irAE hospitalization versus non-irAE hospitalization, are listed in Table 1. Of all patients who were treated with ICIs, 23% were hospitalized for confirmed irAEs. Median age at hospitalization was higher for ICI-treated patients who were admitted for a confirmed irAE versus those not admitted for irAEs (68 years v 59 years; P = .011). Median length of stay for patients with irAE hospitalizations was comparable to patients with non-irAE hospitalizations in the ICI-treated population (median, 6 days v 5 days; P = .7). More male patients were admitted than female patients (65% v 35%) for both irAE- and non–irAE-related admissions. Patients with confirmed irAEs most commonly had thoracic/head and neck cancers (35%), melanoma/skin cancer (39%), and GI cancers (13%; P = .088). Among the 23 patients who were hospitalized with confirmed irAEs, an equal proportion had received anti–PD-1/PD-L1 monotherapy or anti–PD-1/PD-L1–based combinations (48% each). Median number of ICI doses before admission was one for both those who developed and did not develop irAEs.

TABLE 1.

Characteristics of Patients Treated With ICIs

The cohort of ICI-treated patients was also separated by whether treatment was received in clinical trial or as standard of care, depicted in Table 1. Eighteen percent of patients who were treated with ICIs in clinical trials were hospitalized for irAEs, whereas 28% of patients treated with standard-of-care ICIs were hospitalized with irAEs. Those who received ICI therapy in clinical trials versus as standard of care and developed irAEs received comparable doses of ICI therapy (median dose, four v three). In addition, patients who were hospitalized for irAEs in a clinical trial were less likely to have received anti–PD-1/PD-L1–based combinations (22%), whereas those who received standard-of-care ICI and developed an irAE received anti–PD-1/PD-L1 monotherapy (64%).

Of the 23 confirmed irAEs, the majority were high grade (grade ≥ 3) and no patients with grade 1 irAEs were admitted (Common Terminology Criteria for Adverse Events grade 2: 35% v grade ≥ 3: 65%). The most common organ-specific toxicities were pneumonitis (26%) and colitis (17%). Six patients were diagnosed with two or more irAEs each during hospitalization. These combinations included diabetes mellitus type I and colitis, pneumonitis and interstitial nephritis, immune-related fever and myocarditis, colitis and thyroiditis, colitis and fatigue, and hypophysitis and pneumonitis. In these cases, we reported the irAE for which the patient was hospitalized as the second irAE did not warrant hospitalization. Nearly all irAEs in this series prompted medicine subspecialist consultations (91%), of those most commonly gastroenterology/hepatology (22%). Medicine subspecialists were consulted for both low- and high-grade irAEs (grade 2: 88% v grade ≥ 3: 93%), and outpatient subspecialist medicine follow-up was obtained in eight of these patients (38%).

Management and Outcomes of irAEs

Clinical data on irAE management and outcomes are listed in Table 2. There was a trend toward increased intravenous corticosteroid use for patients with grade 3 or higher irAEs versus other management (67%; P = .11). Median duration of corticosteroid use was 34 days (range, 5 to 134 days) for grade 2 toxicity versus 39 days (range, 3 to 68 days) for grade 3 or higher toxicity. Median length of stay was 5 days for patients with grade 2 irAEs and 9 days for those with grade 3 or higher irAEs. Most (87%) patients were discharged after irAE management.

TABLE 2.

Highest Treatment and Clinical Outcome Details for irAEs

Regarding outcomes, 15 patients had their irAEs improve or resolve (65%), seven patients’ irAEs (30%) clinically worsened during hospitalization, and three patients died of their irAE. Causes of death included transplanted renal graft failure, infliximab-refractory pneumonitis, and corticosteroid-refractory interstitial nephritis. Most patients whose irAEs worsened had high-grade irAEs (71% v 29%; P < .001). After discharge, nearly all patients (87%) discontinued immunotherapy.

irAE Readmissions

Eight patients (8%) in the ICI-treated cohort were readmitted for a suspected irAE, three of whom had a confirmed, recurrent episode of the initial irAE at a higher grade on second hospitalization (hypophysitis, n = 1; hepatitis, n = 1; colitis, n = 1). In these cases, readmission occurred between 4 and 8 weeks after the first hospitalization. All three patients who were readmitted for irAEs clinically improved, were treated with fewer than five doses of anti–PD-1 monotherapy (nivolumab), and did not receive ICIs between hospitalizations.

Associations Between Patient Features and irAEs

Associations between patient features and irAE hospitalizations among immunotherapy-treated patients who required admission are depicted in Figure 1. Among hospitalized ICI-treated patients, those who received combination ICIs (odds ratio [OR], 6.8; 95% CI, 2.0 to 23.2) and those older than age 65 years (OR, 5.4; 95% CI, 1.6 to 17.8) were more likely to be hospitalized for an irAE. We observed a trend toward an increased risk of irAE hospitalization in those who had chronic irAEs (OR, 2.3; 95% CI, 0.77 to 7.1). No significant association was observed between prior autoimmune disease and irAE hospitalization (OR, 1.0; 95% CI, 0.3 to 4.0).

DISCUSSION

Here, we report one of the first comprehensive series of patients with solid tumor malignancies who have been hospitalized for irAEs. We observed that nearly one quarter of hospitalized patients with solid tumors in an academic center had received or were actively receiving ICIs for treatment of cancer. The high prevalence of ICI-treated hospitalized oncology patients reflects the growing number of US Food and Drug Administration approvals and currently available ICI-containing clinical trials for cancer.^22-24 The incidence of irAEs in reported late-phase clinical trials ranges from 19.1% to 64%, with high-grade irAEs in which patients would have been hospitalized accounting for less than 5% of cases.^25-27 Our study demonstrates that a notable proportion of ICI-treated patients were hospitalized for confirmed irAEs (23%) across both the clinical trial (18%) and nontrial (28%) settings. Of importance, these patients accounted for 5% of the entire cohort of hospitalized oncology patients—both ICI and non-ICI treated—in accordance with published clinical trial data.^25-27 In this study, we found that the most common irAEs warranting hospital admission were pneumonitis (26%) and colitis (17%). We also observed that the spectrum of toxicities in our series was somewhat different from that of published clinical trials,^27a which may be related to local patterns and irAE expertise. Unsurprisingly, most patients who were hospitalized for irAEs had high-grade irAEs or uncommon irAEs.^28,29 In terms of irAE management, we found that the majority of patients with confirmed irAEs required medical subspecialty consults. In addition, we identified that risk factors for irAE hospitalization included the receipt of combination ICIs and age older than 65 years. Last, we observed that irAE outcomes, despite hospitalization, were generally favorable, with the majority of patients having their irAEs resolve/improve.

To our knowledge, there are limited data on hospitalizations for irAEs outside of our experience. Only one reported abstract describes irAE hospitalizations in another large academic cancer center.³⁰ In that study, the authors reported that 65% of suspected irAEs were confirmed irAEs, which is consistent with our findings; however, a markedly higher readmission rate was observed in this study versus ours (61.7% v 8%), reflecting potential differences in patient populations, subspecialist services, familiarity with specific toxicities, or length of study. In addition, the study in question examined whether oncologists believed irAEs should be managed by medicine rather than oncology services and showed this to be a view shared by 48% of those surveyed. Our study also examined the role of multidisciplinary input, with 91% of those patients with confirmed irAEs undergoing subspecialty consultations.

We identified that the receipt of combination ICIs was associated with an increased likelihood of irAE-related hospitalizations compared with non-irAE hospitalizations of patients with cancer. This is consistent with published data from clinical trials.^25-27 Moreover, we demonstrated that age older than 65 years is a risk factor for irAE-related hospital admission. This may be a result of greater high-grade irAE risk in older patients, or that this population is under-represented in clinical trials. We did not detect an association between prior autoimmune conditions and irAE hospitalizations, and data from published literature have also been conflicting on this topic.^31,32 More studies with larger sample sizes are warranted to further assess the association between autoimmune disease and irAE risk.

Two of three patients who died in this cohort developed corticosteroid-refractory toxicities, namely pneumonitis and nephritis. There are no standard immunosuppressive treatments after corticosteroid failure for either of these irAEs, with guidance provided only by expert consensus.^17-19 Our study therefore highlights the need for prospective and mechanistic studies of patients who develop corticosteroid-refractory irAEs. The third patient in this study died of a solid organ transplantation rejection, which underscores the potentially higher risk of ICIs in this patient population historically which had been excluded from ICI clinical trials.^28,33 In our experience, fatality from irAEs was a rare event, consistent with results from published studies.^25-27

In summary, data from our study provide important initial insight into current hospital services required to care for patients with irAEs and have implications for future service planning. Although not significant, we observed a numerically prolonged length of stay for patients with irAE hospitalizations, particularly for those with high-grade irAEs. We anticipate that these data may be more striking in settings with limited access to oncologists and medicine subspecialists with specific expertise in irAEs. Nearly all patients in this cohort had medicine subspecialist consultations; however, a minority had follow-up with these specialists after discharge. This may be explained, in part, by patient deaths from high-grade toxicity, informal interactions between medical specialists and oncologists during established oncologic follow-up, or increasing comfort levels of oncologists in managing stable irAEs in the outpatient setting. Hospitals and oncology centers may consider conducting internal audits to identify site-specific service needs for patients with irAEs. These data may support the creation of subspecialist outpatient clinics and medicine training programs for organ specialists in irAEs.

Limitations of our study include a small sample size, focus on hospitalized patients alone, limited study timeframe, and the generalizability of our findings to community oncology settings. The patient population in this study is from an academic center and therefore included a notable proportion of patients who were treated in clinical trials. This subset of patients may have an inherently increased risk of irAEs, which may have led to increased rates of hospitalization.^11,34 Conversely, whereas academic centers may see and treat larger numbers of patients with ICIs in clinical trials, the incidence of irAE hospitalization and irAE severity may be higher in community settings because of a lack of local expertise in managing irAEs. To our knowledge, specific data on irAEs in community oncology settings have yet to be published. Regardless, with newly US Food and Drug Administration–approved ICI combinations being used as standard of care in patients with melanoma, non–small-cell lung cancer, and renal-cell carcinoma, observations from patients in this study are likely to become increasingly relevant across all settings. Another potential limitation of our study is related to irAE adjudication, as potential biases may have been introduced through the retrospective nature of confirmation of irAE events, and data abstraction that did not involve organ specialists. Of importance, our study provides a snapshot of inpatient irAEs over a 6-month period and thus may not capture variations in hospitalizations across the full calendar year (Appendix, online only), readmission rates over a longer period, and irAEs diagnosed and managed in the outpatient setting. Last, whereas we observed frequent multidisciplinary care for patients with irAEs in our cohort, future studies will need to evaluate the objective benefit of this approach.

In conclusion, to our knowledge, this is one of the first studies to report the proportion of oncology patients hospitalized for suspected and confirmed irAEs. We found that older patients and those being treated with ICI combinations are at higher risk for irAE-related admission. Crucially, these data may have important implications for current and future service provision on inpatient oncology units, including multidisciplinary care, for patients hospitalized for irAEs.

APPENDIX

Variations in medical oncology and confirmed immune-related adverse event (irAE) admissions were separated by month. In May, there were 22 medical admissions and 0 irAE admissions; June: 74 medical oncology admissions and seven irAE admissions; July: 90 medical oncology admissions and two irAE admissions; August: 59 medical oncology admissions and four irAE admissions; September: 43 medical oncology admissions and one irAE admission; October: 62 medical oncology admissions and four irAE admissions; November: 68 medical oncology admissions and four irAE admissions; and December: 25 medical oncology admissions and one irAE admission. Admissions for the month of May were truncated as data collection began only on May 30, 2017; therefore, admissions that were discharged before May 30, 2017, were not included in our study. In observing the data, there were no discernable trends in either medical oncology or irAE admissions.

Fig A1.

CONSORT diagram of admissions to the inpatient oncology unit over 6 months. Inpatient patients were stratified on the basis of whether patients received immune checkpoint inhibitors, their hospital workup included suspected immune-related adverse event (irAE), and they subsequently had a confirmed irAE during their hospital stay.

AUTHOR CONTRIBUTIONS

Conception and design: Aanika Balaji, Jiajia Zhang, Jarushka Naidoo

Financial support: Jarushka Naidoo

Provision of study materials or patients: Jarushka Naidoo

Collection and assembly of data: Aanika Balaji, Jiajia Zhang, Hany Elmariah, Jacquelyn W. Zimmerman, Won Jin Ho, Joseph Heng, Paz Vellanki, Jarushka Naidoo

Data analysis and interpretation: Aanika Balaji, Jiajia Zhang, Kristen A. Marrone, Mark Yarchoan, Deepti Venkatraman, Won Jin Ho, Joshua E. Reuss, Matthias Holdhoff

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Immune-Related Adverse Events Requiring Hospitalization: Spectrum of Toxicity, Treatment, and Outcomes

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jop/site/ifc/journal-policies.html.

Kristen A. Marrone

Consulting or Advisory Role: AstraZeneca, Compugen

Mark Yarchoan

Consulting or Advisory Role: Eisai, Exelixis

Research Funding: Bristol-Myers Squibb (Inst), Merck (Inst), Exelixis (Inst)

Deborah K. Armstrong

Consulting or Advisory Role: CUE Biopharma

Research Funding: Clovis Oncology (Inst), AstraZeneca (Inst), Advaxis (Inst), Syndax (Inst), Pfizer (Inst), Tesaro (Inst)

Daniel A. Laheru

Patents, Royalties, Other Intellectual Property: Colon GVAX

Ranee Mehra

Stock and Other Ownership Interests: GlaxoSmithKline (I)

Consulting or Advisory Role: Genentech, Bayer

Research Funding: AstraZeneca

Won Jin Ho

Patents, Royalties, Other Intellectual Property: Patents and royalties from Rodeo Therapeutics

Matthias Holdhoff

Consulting or Advisory Role: Celgene, AbbVie, BTG, Newlink Genetics, DPClinical

Travel, Accommodations, Expenses: Arbor Pharmaceuticals

Jarushka Naidoo

Honoraria: Bristol-Myers Squibb, AstraZeneca, MedImmune

Consulting or Advisory Role: Bristol-Myers Squibb, AstraZeneca, MedImmune, Genentech

Research Funding: Merck (Inst), Calithera Biosciences (Inst), AstraZeneca (Inst)

Travel, Accommodations, Expenses: Bristol-Myers Squibb, AstraZeneca, MedImmune

No other potential conflicts of interest were reported.