Volume 126, Issue 16 p. 3758-3767
Original Article
Free Access

Vitamin D intake is associated with decreased risk of immune checkpoint inhibitor-induced colitis

Shilpa Grover MD, MPH

Shilpa Grover MD, MPH

Harvard Medical School, Boston, Massachusetts

Division of Gastroenterology, Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts

Search for more papers by this author
Michael Dougan MD, PhD

Michael Dougan MD, PhD

Harvard Medical School, Boston, Massachusetts

Division of Gastroenterology, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts

Search for more papers by this author
Kevin Tyan BA

Kevin Tyan BA

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Anita Giobbie-Hurder MS

Anita Giobbie-Hurder MS

Division of Biostatistics, Department of Data Sciences, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Steven M. Blum MD

Steven M. Blum MD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Broad Institute at MIT and Harvard, Cambridge, Massachusetts

Massachusetts General Hospital Cancer Center, Boston, Massachusetts

Search for more papers by this author
Jeffrey Ishizuka MD, DPhil

Jeffrey Ishizuka MD, DPhil

Departments of Medicine and Pathology, Yale University School of Medicine, New Haven, Connecticut

Search for more papers by this author
Taha Qazi MD

Taha Qazi MD

Division of Gastroenterology, Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts

Digestive Disease & Surgery Institute, Cleveland Clinic, Cleveland, Ohio

Search for more papers by this author
Rawad Elias MD

Rawad Elias MD

Hartford HealthCare Cancer Institute, Hartford, Connecticut

Search for more papers by this author
Kruti B. Vora BA

Kruti B. Vora BA

Harvard Medical School, Boston, Massachusetts

Search for more papers by this author
Alex B. Ruan MD

Alex B. Ruan MD

Harvard Medical School, Boston, Massachusetts

Search for more papers by this author
William Martin-Doyle MD, MPH

William Martin-Doyle MD, MPH

Harvard Medical School, Boston, Massachusetts

Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts

Search for more papers by this author
Michael Manos BA

Michael Manos BA

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Lauren Eastman BA

Lauren Eastman BA

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Meredith Davis MS, PA-C

Meredith Davis MS, PA-C

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Maria Gargano MS, PA-C

Maria Gargano MS, PA-C

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Rizwan Haq MD, PhD

Rizwan Haq MD, PhD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Elizabeth I. Buchbinder MD

Elizabeth I. Buchbinder MD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts

Search for more papers by this author
Ryan J. Sullivan MD

Ryan J. Sullivan MD

Massachusetts General Hospital Cancer Center, Boston, Massachusetts

Search for more papers by this author
Patrick A. Ott MD, PhD

Patrick A. Ott MD, PhD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Broad Institute at MIT and Harvard, Cambridge, Massachusetts

Department of Medicine, Brigham and Women's Hospital, Boston, Massachusetts

Search for more papers by this author
F. Stephen Hodi MD

F. Stephen Hodi MD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Search for more papers by this author
Osama E. Rahma MD

Corresponding Author

Osama E. Rahma MD

Harvard Medical School, Boston, Massachusetts

Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

Corresponding Author: Osama E. Rahma, MD, Center for Immuno-Oncology, Gastrointestinal Cancer Center, Dana-Farber Cancer Institute, Harvard Medical School, 450 Brookline Ave, Boston MA 02215 ([email protected])

Search for more papers by this author
First published: 22 June 2020
Citations: 36
The first 2 authors contributed equally to this article.
Patients evaluated in this study were identified from the Dana-Farber Cancer Institute melanoma bio-specimen banking protocol (DFCI IRB approved protocol 05-042) and Massachusetts General Hospital (MGH IRB approved protocol 2017P002740/PHS).

Abstract

Background

There is a lack of predictive markers informing on the risk of colitis in patients treated with immune checkpoint inhibitors (ICIs). The aim of this study was to identify potential factors associated with development of ICI colitis.

Methods

We performed a retrospective analysis of melanoma patients at Dana-Farber Cancer Institute who received PD-1, CTLA-4, or combination ICIs between May 2011 to October 2017. Clinical and laboratory characteristics associated with pathologically confirmed ICI colitis were evaluated using multivariable logistic regression analyses. External confirmation was performed on an independent cohort from Massachusetts General Hospital.

Results

The discovery cohort included 213 patients of whom 37 developed ICI colitis (17%). Vitamin D use was recorded in 66/213 patients (31%) before starting ICIs. In multivariable regression analysis, vitamin D use conferred significantly reduced odds of developing ICI colitis (OR 0.35, 95% CI 0.1–0.9). These results were also demonstrated in the confirmatory cohort (OR 0.46, 95% CI 0.2–0.9) of 169 patients of whom 49 developed ICI colitis (29%). Pre-treatment neutrophil-to-lymphocyte ratio (NLR) ≥5 predicted reduced odds of colitis (OR 0.34, 95% CI 0.1–0.9) only in the discovery cohort.

Conclusions

This is the first study to report that among patients treated with ICIs, vitamin D intake is associated with reduced risk for ICI colitis. This finding is consistent with prior reports of prophylactic use of vitamin D in ulcerative colitis and graft-versus-host-disease. This observation should be validated prospectively in future studies.

Background

Immune checkpoint inhibitors (ICIs) targeting cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) and the programmed cell death-1 (PD-1)/PD-ligand 1 (PD-L1) pathway have improved the prognosis for many malignancies.1, 2 The anti-tumor activity of these agents results from augmentation of the T-cell immune response by blocking two immune checkpoints, CTLA-4 and PD-1/PD-L1 that protect against detrimental inflammation and autoimmunity.3, 4 This augmented immune response can also lead to a range of systemic and organ specific immune-related adverse effects (irAE).5 The burden of irAEs hinders the clinical benefit of ICIs, with over 50% of patients on combination immunotherapy developing grade 3-4 toxicities.6, 7 In particular, immune-related colitis is among the most frequent and severe irAEs often leading to treatment interruption, immunosuppression that may negatively impact tumor immune response, and ICI discontinuation.8-11 The incidence of diarrhea and colitis is higher using CTLA-4-blocking agents compared to PD1/PD-L1 blocking agents,6, 12 and highest with a combination of anti-CTLA-4 and anti-PD-1 antibodies.11, 13

ICI-induced colitis shares similar phenotypical, histological, and serological characteristics with inflammatory bowel disease (Crohn's disease [CD] and ulcerative colitis [UC]), albeit with a more aggressive disease course involving acute rather than chronic inflammation.11, 14, 15 Several individual (e.g., male gender) and environmental (e.g., smoking, vitamin D deficiency, NSAID use, PPI use) risk factors have been associated with the risk of IBD or its complications.2, 16-21 High vitamin D levels are a known protective factor for both CD and UC.18 A range of comorbidities, likely due to a common pathogenesis of immune dysregulation, have also been shown to correlate with IBD, including history of autoimmune disease22, 23 and eosinophilia.24, 25 There are comparatively few predictive markers for determining ICI-induced colitis risk.

In this retrospective cohort study, we reviewed potential factors that correlate with grade 3-4 irAEs in melanoma patients who received ICIs at Dana-Farber Cancer Institute (DFCI). We found an association between lower levels of albumin prior to treatment with ICIs and the development of high-grade irAEs.26 Subsequently, we performed a secondary analysis to identify factors that are associated with specific irAEs. Here we report findings pertaining to ICI-induced colitis in our DFCI cohort. These results were independently confirmed in a cohort of melanoma patients treated with ICIs at Massachusetts General Hospital (MGH).

Methods

Discovery Cohort Patient Population and Characteristics

246 patients included in the discovery cohort of this study were identified from DFCI's melanoma bio-specimen banking protocol (DFCI protocol 05-042). Patients in this cohort received at least one course of pembrolizumab, nivolumab, or ipilimumab alone, or the combination of ipilimumab and nivolumab between May 2011 to October 2017. Patient characteristics and concomitant medications were collected from the electronic medical record (EMR) at the time of the first ICI treatment initiation. Pre-treatment use of vitamin D was evaluated based on reports of vitamin D intake at the time the first ICI treatment was initiated, including therapeutic vitamin D (ergocalciferol and cholecalciferol) and multivitamin use containing at least 400 IU of vitamin D. Vitamin D intake was divided into 3 dosage categories: no use, ≤1000 IU, and >1000 IU per day; selection of the 1000 IU dose stratification was based on standard guidelines for recommended daily supplementation and repletion.27, 28

Other laboratory values (albumin, lactate dehydrogenase [LDH], neutrophil-to-lymphocyte ratio [NLR], and eosinophils) were collected at the time of the treatment initiation for a given line of therapy. Peri-treatment vaccination history with influenza and pneumococcal pneumonia vaccines was determined based on documentation of immunization within 3 months before or after initiating ICI therapy. Infections (pneumonia, bacteremia, UTI, or sepsis) within 6 months of initiating ICIs were noted. Comorbidities including autoimmune disease, asthma, or seasonal allergies were recorded. Tumor characteristics, including prior therapy within 6 months of the ICIs initiation, common tumor mutations (c-KIT, BRAF, NRAS), and number of metastatic sites were collected. The protocol was approved by the DFCI Institutional Review Board.

Outcomes

The primary outcome was the development of immune-related colitis as confirmed by histopathological biopsies of the colon. In the discovery cohort, all patients with grade 3-4 diarrhea were evaluated by flexible sigmoidoscopy and immune-related colitis was confirmed by histopathological biopsy. Patients with negative biopsies were included in the final cohort analysis as non-colitis cases. irAE grading was determined by review of the EMR. If specific grades were not documented in the chart by clinicians, then a single reviewer graded the irAEs using the NIH Common Terminology Criteria for Adverse Events (CTCAE) Version 4.03. Data abstraction on the DFCI cohort was concluded by October 2017. irAEs were reviewed by a second reviewer to confirm the events.

Confirmatory Cohort

External confirmation was performed on an independent cohort of 169 melanoma patients treated with PD-1, CTLA-4 inhibitors or the combination at MGH from December 2010 until July 2018 under IRB approved protocol (2017P002740/PHS). Cases of immune-related colitis were identified by evaluating all patients with grade 3-4 diarrhea by flexible sigmoidoscopy and confirmation with histopathological biopsy. Patients with negative biopsies were not included in the final cohort analysis. The non-colitis patients in this cohort did not report gastrointestinal symptoms and did not undergo endoscopy or biopsy.

Statistical Methods

Univariate analysis was used to evaluate the relationships between the development of ICI-induced colitis and patient and clinical covariates of interest. Fisher's exact tests were used for comparisons of categorical variables and Wilcoxon rank-sum tests for continuous variables. A multivariable prediction model for ICI-induced colitis was constructed using logistic regression. Since patients could have received more than one checkpoint therapy, ICI treatment and irAE data relevant to that ICI treatment were correlated within each patient. To address this, a multivariable, generalized linear model was fit to the data, allowing for multiple occurrences of irAEs on different ICI therapies. The binary outcome variable was modeled using a logit link function with a binomial distribution. A multivariable regression model of immune-related colitis was fit using characteristics that had a nominal univariate Fisher's exact P-value of .30 or less as candidate predictors. Forward, backward, and stepwise elimination were used to assure model consistency. Predictors with a P-value of .05 or less were selected for the final model. For the confirmatory cohort, ICI class, neutrophil-to-lymphocyte ratio divided at 5, and vitamin D use were the only independent predictors in the logistic regression. Results are reported as odds ratios (OR) with 95% Wald confidence intervals. All reported P-values are 2-sided. Statistical analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

Discovery Cohort Characteristics and Incidence of Immune-Related Adverse Events

We assessed the incidence of irAEs in 246 patients with stage III or IV melanoma treated with ICIs at DFCI from May 2011 to October 2017 (Fig. 1). We excluded 23 patients for lack of irAEs, then limited the dataset to standard-of-care ICIs that were FDA approved at the time of the analysis, which further reduced the dataset to 213 patients. Of these 213 patients included in the analysis, 49 patients received more than one ICI regimen during their course of treatment, resulting in 267 individual ICI treatment regimens. 164 patients received a single regimen, 44 had two, and 5 had three regimens (two PD-1 and one CTLA-4 inhibitor). Of the 267 ICI regimens, 66 were ipilimumab, 17 nivolumab, 91 pembrolizumab, and 93 combined ipilimumab and nivolumab (Table 1). The most common toxicities were dermatitis and colitis (17.8% each), followed by rheumatologic (12.7%), endocrine (10.3%), hepatitis (7.5%), and pneumonitis (7.5%, Supporting Table 1).

Details are in the caption following the image
Study schema for patients included in discovery cohort analysis.
Table 1. Immune-Checkpoint Inhibitor Regimen at Time of Highest Grade irAE in the Discovery Cohort
Maximum irAE Grade, No.
1 or 2 3 or 4
Ipilimumab and nivolumab 30 63
Ipilimumab 33 33
Nivolumab 12 5
Pembrolizumab 59 32
  • Abbreviation: irAE, immune-related adverse effect.
  • This table categorizes the commercially available immunotherapy treatment that patients in the discovery cohort were taking at the time of their highest grade irAE, delineated by grades 1-2 vs. grades 3-4.

Among the 213 individual patients in the discovery cohort, 70 patients reported gastrointestinal symptoms, of which 41 had high-grade diarrhea that were evaluated by flexible sigmoidoscopy and biopsy. There were 38 reports of pathologically confirmed ICI-induced colitis in 37 patients (17.4%, Table 2); 8 presentations were grade 2, 27 were grade 3, and 3 were grade 4 colitis. Four patients had negative biopsies for colitis and were included in the final cohort analysis. The majority of colitis cases (32/38, 84%) occurred during therapy with ipilimumab monotherapy (17/38, 45%) or during combined therapy of nivolumab and ipilimumab (15/38, 39%); the rest occurred during nivolumab or pembrolizumab therapy (6/38, 16%, Table 2).

Table 2. Incidence of Colitis Based on Immune Checkpoint Inhibitor Regimen in Discovery and Confirmatory Cohorts
Checkpoint Class Colitis in Discovery Cohort, No. (%) Colitis in Validation Cohort, No. (%)
No (n = 229) Yes (n = 38) No (n = 196) Yes (n = 50)
Combination (ipilimumab + nivolumab) 78 (34.1) 15 (39.5) 22 (11.2) 11 (22.0)
Ipilimumab 49 (21.4) 17 (44.7) 61 (31.2) 23 (46.0)
Nivolumab 16 (7.0) 1 (2.6) 14 (7.1) 3 (6.0)
Pembrolizumab 86 (37.6) 5 (13.2) 99 (50.5) 13 (26.0)
  • This table summarizes the incidence of immune-related colitis based on ICI regimen in the discovery cohort of 213 patients receiving a total of 267 regimens, and the confirmatory cohort of 169 patients receiving 246 regimens.

Association of Clinical Characteristics With ICI Colitis

We performed univariate analysis to evaluate the association of development of ICI colitis and clinical characteristics (demographics, comorbidities, peri-treatment immunization, pre-treatment laboratory values, tumor characteristics, and concomitant medications, Table 3). Vitamin D use was recorded in 66/213 patients (31%, Table 4) before starting ICIs and showed a statistically significant association with decreased incidence of ICI colitis (34.1% of patients without colitis reported vitamin D intake vs. 16.2% of patients with colitis, P = .03). Of the 66 patients taking vitamin D, 30 reported dosages of 1000 IU or less, and 36 had dosages greater than 1000 IU.

Table 3. Patient Characteristics, Comorbid Disease History, Pre-treatment Laboratory Values, Disease and Tumor Characteristics, and Concomitant Medications in the Discovery Cohort
Colitis, No. (%) Fisher Exact P
No (n = 176) Yes (n = 37)
Patient characteristics
Male 102 (58.0) 22 (59.5) .99
Age < 70 y 117 (66.5) 21 (56.8) .26a
Obese (BMI ≥ 30 kg/m2) 43 (24.4) 7 (18.9) .53
Pretreatment ECOG PS ≥ 1 34 (19.3) 5 (13.5) .49
History of smoking 80 (45.5) 18 (48.6) .72
History of alcohol use 118 (67.0) 24 (64.9) .85
Influenza immunization within 3 mo of treatment 34 (19.3) 12 (32.4) .08a
Pneumococcal pneumonia immunization within 3 mo of treatment 35 (19.9) 13 (35.1) .05a
Influenza or pneumonia immunization within 3 mo of treatment 50 (28.4) 17 (45.9) .05a
Comorbid diseases
Infection in previous 6 mo 13 (7.4) 3 (8.1) .99
Autoimmune disease 12 (6.8) 2 (5.4) .99
Asthma 17 (9.7) 2 (5.4) .54
Seasonal allergies 20 (11.4) 4 (10.8) .99
Asthma or seasonal allergies 32 (18.2) 6 (16.2) .99
Pretreatment laboratory values
Albumin ≤ 4.2 g/dL 96 (54.5) 22 (59.5) .72
Normal LDH (110-220 U/L) 110 (62.5) 26 (70.3) .45
NLR ≥ 5 55 (31.3)b 6 (16.2) .16a
NLR ≥ 3 107 (60.8) 18 (48.6) .34
Absolute eosinophil count > 400/µL 29 (16.5)c 5 (13.5) .81
Cancer: disease and treatment characteristics
Prior chemotherapy 34 (19.3) 7 (18.9) .99
Prior targeted therapy 37 (21.0) 5 (13.5) .37
Prior chemotherapy or targeted therapy within previous 6 mo 10 (5.7) 0 (0) .22a
Prior radiation therapy in previous 6 months 44 (25.0) 12 (32.4) .41
c-KIT mutation 5 (2.8) 1 (2.7) .99
BRAF mutation 49 (27.8) 9 (24.3) .84
NRAS mutation 26 (14.8) 4 (10.8) .79
c-KIT, BRAF, or NRAS mutation 77 (43.8) 14 (37.8) .59
No. of metastatic sites > 1 174 (98.9) 37 (100) .99
No. of metastatic sites > 3 85 (48.3) 21 (56.8) .37
Concomitant medications
ACEi 29 (16.5) 7 (18.9) .81
ACEi or ARB 41 (23.3) 10 (27.0) .67
Aspirin 29 (16.5) 8 (21.6) .48
Aspirin or NSAID 38 (21.6) 11 (29.7) .29a
Proton-pump inhibitor 42 (23.9) 6 (16.2) .39
Statin 39 (22.2) 10 (27.0) .52
Vitamin D supplementation 60 (34.1) 6 (16.2) .03a
  • Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BMI, body mass index; ECOG, Eastern Cooperative Oncology Group; LDH, lactate dehydrogenase; NLR, neutrophil-to-lymphocyte ratio; NSAID, nonsteroidal anti-inflammatory drug; PS, performance status.
  • This table summarizes the characteristics of 213 patients based upon those who developed ICI colitis vs. those who did not develop colitis.
  • a P < .30, which met inclusion for consideration in the multivariable model.
  • b Missing NLR values from 1 colitis and 7 non-colitis patients.
  • c Missing eosinophil count from 1 colitis and 3 non-colitis patients.
Table 4. Incidence of Colitis Based on Vitamin D Intake in Discovery and Confirmatory Cohorts
Colitis in Discovery Cohort, N (%) Colitis in Confirmatory Cohort, N (%)
No (N = 176) Yes (N = 37) No (N = 120) Yes (N = 49)
Vitamin D Intake
Yes 60 (34.1) 6 (16.2) 80 (66.7) 23 (46.9)
No 116 (65.9) 31 (83.8) 40 (33.3) 26 (53.1)
  • This table summarizes the incidence of immune-related colitis based on vitamin D intake initiated before ICI therapy in the discovery cohort of 213 patients and the confirmatory cohort of 169 patients.

Age divided at 70, peri-treatment pneumococcal or influenza immunization, NLR ≥ 5, prior chemotherapy and/or targeted therapy, and concomitant aspirin or NSAIDs met the univariate P-value threshold of .30 for potential association with increased risk of ICI colitis and were thus included in the multivariable regression model as candidate predictors.

Certain factors that were reviewed, including demographics, comorbidities, pre-treatment laboratory values, tumor characteristics, prior treatment, and concomitant medications, were not found to demonstrate statistically significant association with the development of ICI colitis (Table 3).

Multivariable Model of Predictors for ICI Colitis

Multivariable regression revealed three independent predictors of immune-related colitis: checkpoint inhibitor class, baseline NLR ≥ 5, and vitamin D use (Table 5).

Table 5. Multivariable Logistic Regression Analysis of Risk Factors for Developing Colitis
Discovery Cohort (N = 213) Confirmatory Cohort (N = 169)
Odds Ratio 95% CI P value Odds Ratio 95% CI P value
Immune Checkpoint Inhibitor Class
Ipi + Nivo vs. Pembrolizumab 3.34 1.1-9.8 .001 3.37 1.4-8.1 .009
Nivolumab vs. Pembrolizumab 1.04 0.1-9.3 1.25 0.3-5.3
Ipilimumab vs. Pembrolizumab 7.48 2.6-21.8 2.68 1.5-4.9
CTLA-4 (Ipi) vs. PD-1 (Pembro or Nivo) 7.14 2.2-25.0 2.38 1.04-5.6
Ipi + Nivo vs. Ipilimumab 0.45 0.2-1.0 1.26 0.5-2.9
Neutrophil-to-Lymphocyte Ratio
≥5 vs. <5 0.34 0.1-0.9 .046 0.61 0.3-1.3 .38
Vitamin D Intake
Yes vs. No 0.35 0.1-0.9 .01 0.46 0.2-0.9 .03
  • Abbreviation: CI, confidence interval.
  • Characteristics with a univariate Fisher's exact P-value of .30 or less were incorporated into a multivariable logistic regression. The model demonstrated increased odds of developing colitis in patients treated with combination (ipilimumab + nivolumab) or ipilimumab monotherapy compared to those treated with pembrolizumab monotherapy. In addition, CTLA-4 monotherapy (ipilimumab) led to increased odds of developing colitis when compared to the pooled PD-1 monotherapy group (pembrolizumab or nivolumab). Furthermore, NLR ≥ 5 and pre-treatment vitamin D supplementation were associated with decreased odds of colitis. The confirmatory cohort confirmed the relationship between colitis and vitamin D intake.

Treatment with combination therapy (ipilimumab and nivolumab) was associated with significantly increased odds of developing ICI colitis compared to treatment with pembrolizumab monotherapy (OR 3.34, 95% CI 1.1–9.8). In addition, treatment with ipilimumab alone conferred significantly increased odds of colitis compared to pembrolizumab monotherapy (OR 7.48, 95% CI 2.6–21.8). Similarly, CLTA-4 monotherapy (ipilimumab) demonstrated increased odds of colitis compared to a pooled PD-1 monotherapy group of pembrolizumab or nivolumab (OR 7.14, 95% CI 2.2–25.0). There was no statistically significant difference in the risk of developing ICI colitis between patients treated with nivolumab monotherapy compared to pembrolizumab monotherapy (OR 1.04, 95% CI 0.1–9.3).

Notably, patients taking vitamin D at the time their first ICI treatment was initiated had significantly reduced odds of developing colitis compared with patients who were not on pre-treatment vitamin D (OR 0.35, 95% CI 0.1–0.9, P = .01). Furthermore, pre-treatment NLR ≥ 5 was also associated with reduced odds of colitis (OR 0.34, 95% CI 0.1–0.9, P = .046).

Given the association of vitamin D with decreased risk of colitis, we compared different dosages of vitamin D intake and their association with ICI colitis. In the refit multivariable regression model, there was no statistically significant difference in the risk of developing ICI colitis between patients taking >1000 IU of vitamin D compared to patients taking ≤1000 IU.

Confirmatory Cohort Analysis

External confirmation of our multivariable model was performed on an independent cohort from MGH, which included 169 melanoma patients treated with 246 regimens of ipilimumab, nivolumab, or pembrolizumab alone, or the combination of ipilimumab and nivolumab. Patients were identified as developing colitis or not developing colitis, and patients were included regardless of whether or not they developed other irAEs. During identification of colitis cases in the confirmatory cohort, 73 patients were found to have high-grade diarrhea and were evaluated by flexible sigmoidoscopy and biopsy. There were 50 histopathologically confirmed reports of immune-related colitis in 49 of the 169 patients (29.9%); 22 presentations were grade 2, 15 cases were grade 3, and none were grade 4 colitis. The remaining 24 patients with high-grade diarrhea but negative biopsies were not included in the final cohort analysis. Thirty-four of the 50 cases of colitis occurred during ipilimumab monotherapy or during combined therapy of ipilimumab with nivolumab (Table 2). Vitamin D use at the time of first ICI treatment initiation was reported in 103 of the 169 patients (60.9%, Table 4), or 137 of the 246 ICI treatment regimens. 66.7% of patients without gastrointestinal symptoms or colitis reported vitamin D intake vs. 46.9% of patients with colitis.

Multivariable regression analysis of the confirmatory cohort confirmed our finding of an association between vitamin D intake and decreased odds of ICI colitis (OR 0.46, 95% CI 0.2–0.9, P = .03). However, the confirmatory cohort did not confirm an association between NLR ≥ 5 and decreased risk for ICI colitis (OR 0.61, 95% CI 0.3–1.3, P = .38).

Patients in the confirmatory cohort treated with combination therapy had increased odds of developing colitis compared to patients on pembrolizumab alone (OR 3.37, 95% CI 1.4–8.1, Table 5). Similarly, patients in the confirmatory cohort treated with ipilimumab monotherapy were more likely to develop colitis compared to patients treated with pembrolizumab alone (OR 2.68, 95% CI 1.5–4.9). CTLA-4 monotherapy (ipilimumab) conferred increased odds of developing colitis compared to patients treated with PD-1 (pembrolizumab or nivolumab) monotherapy (OR 2.38, 95% CI 1.04–5.6). We did not find a statistically significant difference in the risk of developing ICI colitis between patients treated with nivolumab monotherapy compared to pembrolizumab monotherapy (OR 1.25, 95% CI 0.3–5.3).

Discussion

Our study analyzed patient demographic characteristics, medical comorbidities, tumor characteristics, prior therapy, pre-treatment laboratory values, and concomitant medications to identify factors associated with the development of immune-related colitis. We found that pre-treatment vitamin D intake significantly lowered the odds of developing ICI colitis in advanced melanoma patients and confirmed these results in an independent cohort of melanoma patients.29 To our knowledge, this is the first study that identifies vitamin D as a protective factor against the development of ICI colitis.

Prior studies in patients with checkpoint inhibitor colitis have suggested NSAID use,15 pre-existing IBD,30-33 baseline microbiota (enriched in Firmicutes and poor in Bacteroidetes)34 as putative risk factors for the development of ICI colitis. Tumor histology has also been proposed as a risk factor for ICI colitis (melanoma as compared to non-small cell lung carcinoma and renal cell carcinoma) although these findings may be confounded by the disproportionate use of ipilimumab and awareness of colitis risk in melanoma patients.35, 36 Various predictive biomarkers have also been studied, with higher serum interleukin 17 (IL-17) levels at baseline shown to correlate with development of grade 3 colitis,37 and increase in absolute eosinophil count by week 4 of therapy correlated with any irAE.38 However, the role of vitamin D in immune checkpoint inhibitor-induced colitis has not been previously reported.

There are compelling data to suggest that vitamin D plays an important role in the risk of autoimmune disorder exacerbation39-41 and more specifically IBD.42, 43 Vitamin D has immunomodulatory properties by decreasing the pro-inflammatory Th1 response and increasing anti-inflammatory Th2 cells.44 In mouse models, vitamin D prevents autoimmune disorders by suppression of IL-17A and Th17 cells.45, 46 In addition, vitamin D may also promote self-tolerance by inhibiting the differentiation and maturation of dendritic cells and increasing T regulatory cells.47, 48 In the context of its immunomodulatory effects, vitamin D supplementation has been shown to have both prophylactic49-51 and therapeutic52 effects for graft-versus-host-disease. In terms of the involvement of vitamin D in gut mucosal health, studies also suggest that vitamin D has a role in Toll-like receptor activation of macrophages, which triggers antimicrobial activity against intracellular bacteria, and gut vitamin D receptor (VDR) signaling, which inhibits colitis by protecting the epithelial mucosal barrier.53, 54 Polymorphisms in the VDR gene are also associated with susceptibility to inflammatory bowel disease.55 There is growing evidence that vitamin D supplementation reduces disease severity and risk of relapse in IBD patients.56-59

In the multivariable regression model of our discovery cohort, we also found that elevated baseline NLR ≥ 5 was negatively correlated with risk of developing ICI colitis. This finding is similar to that of a recent study in South Korea reporting lower risk of irAEs in patients with baseline NLR ≥ 3 who were treated with pembrolizumab, although this study did not examine colitis specifically.60 Neutrophil-to-lymphocyte ratio is an increasingly used biomarker of systemic inflammation, but there are conflicting reports around its prognostic value in IBD. While some studies have shown that high NLR predicts increased severity of active UC and CD,61, 62 low NLR has also been found to be strongly associated with risk of UC relapse in patients receiving tacrolimus therapy.63 In our study, NLR evaluated at the threshold of 3 did not yield statistically significant associations with ICI colitis. In addition, our initial finding of decreased risk of ICI colitis in patients with NLR ≥ 5 was not confirmed in the confirmatory cohort. Further studies on NLR as a predictive biomarker for irAEs and immune-related colitis are warranted using a larger sample size.

We acknowledge that our study has key limitations, including its retrospective nature, the fact that it only included melanoma patients, and that we evaluated vitamin D intake before the initiation of the first ICI rather than each ICI. The heterogeneity in study populations is both a weakness and strength. Patients without irAEs were not included in the discovery cohort; however, those without any irAEs were a small percentage of the patients. It should be noted that the confirmatory cohort from MGH included a higher percentage of patients with ICI colitis and included a proportion of patients that did not develop any irAEs. The differences between these cohorts may reflect heterogeneity of the patient population or cross-institutional variability in irAE grading. Another key difference between cohorts is that non-colitis patients in the discovery cohort include those with high-grade diarrhea but negative biopsies, whereas the confirmatory cohort only included non-colitis patients who did not report any gastrointestinal symptoms nor underwent biopsies.

We were also limited by sample size for some of the patient characteristics evaluated (medical comorbidities, tumor mutation status, prior infection, and concomitant medication use). The limited number of exposed patients may prevent meaningful analysis of these factors. In particular, the subgroup analysis comparing different vitamin D dosages and association with ICI colitis is limited due to the small sample size of each dosage category, while the skew of ICI colitis towards higher grades prevented meaningful analysis of association between vitamin D intake and ICI colitis severity. The stratification of dosages at 1000 IU was based on clinical guidelines for supplementation and repletion as well as attaining sufficient sample size of each subgroup for statistical analysis. Future studies should include larger cohort sizes to allow for analysis of additional dosage thresholds, while investigating whether vitamin D intake attenuates the grade of colitis.

We were unable to analyze baseline serum levels of vitamin D in patients due to limited serum samples, and chart review of serum vitamin D levels collected during routine clinical evaluation did not yield sufficient baseline timepoints for analysis. Baseline serum vitamin D may be a more meaningful measure of physiologic vitamin D levels compared to vitamin D intake as a proxy, although vitamin D supplementation and serum levels have been shown to follow an exponential dose response curve.64 Patients reporting high dosages of vitamin D intake (>1000 IU) may represent a confounding factor due to being prescribed for underlying disease processes (vitamin D deficiency, osteoporosis, etc.). Future studies should seek to investigate a relationship between specific dosage levels of vitamin D and decreased risk for ICI colitis, as well as confirming results through baseline vitamin D serum levels.

In conclusion, here we report that patients taking vitamin D at the time their first ICI treatment was initiated had significantly reduced odds of developing ICI colitis over the course of their immunotherapy regimen. These results may suggest benefit in prophylactic use of vitamin D supplementation to prevent ICI colitis, as previously demonstrated in inflammatory bowel disease56-59, 65, 66 and graft-versus-host disease.49, 67, 68 The specific mechanism by which vitamin D may prevent immune-related colitis should be further explored through future correlative studies, including cytokine analyses and immune profiling at baseline and at the time of colitis. The results from this retrospective study should be validated prospectively in larger cohorts, which may lead to a better understanding of factors that drive and prevent the development of immune checkpoint inhibitor-induced colitis.

Funding Support

This project is supported by the Parker Institute for Cancer Immunotherapy. Michael Dougan is supported by an American Gastroenterological Association Research Scholars Award and a National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases Mentored Clinical Scientist Training Grant (1K08 DK114563-01). Steven M. Blum is supported by the Massachusetts General Hospital T32 (2T32CA071345-21A1).

Conflict of Interest Disclosures

Shilpa Grover is employed as a senior physician editor for UpToDate, a Wolters Kluwer company. Michael Dougan reports grant support from Novartis and Genentech; is a member of the scientific advisory board for Neoleukin Therapeutics; and has been a consultant for Tillotts Pharma, Genentech, and Partner Therapeutics. Steven M. Blum reports consulting with Two River Consulting and Third Rock Ventures as well as equity in Kronos Bio and Allogene Therapeutics. Jeffrey Ishizuka reports consulting for Two River, Trammel Therapeutics, and Tango Therapeutics and owns equity in Jounce Therapeutics and Kronos Therapeutics. Rizwan Haq has research grant support from Novartis and Bristol-Myers Squibb as well as a consulting arrangement with Tango Therapeutics. Elizabeth I. Buchbinder has served on advisory boards for Array Biopharma, Bristol-Myers Squibb, Trieza Therapeutics, and Novartis, and she also receives clinical trial support from Eli Lilly, Novartis, Bristol-Myers Squibb, Genentech, and BVD. Ryan J. Sullivan reports advisory roles for Array Biopharma, Asana, Amgen, Bristol-Myers Squibb, Novartis, Genentech, Merck, Replimune, and Syndax and grants from Amgen and Merck. Patrick A. Ott reports the following: advisory roles for Alexion, Array, Bristol-Myers Squibb, Celldex, CytomX, Genentech, Merck, Neon Therapeutics, Novartis, Pfizer, and TRM Oncology; institutional grants from Armo Biosciences, AstraZeneca/MedImmune, Bristol-Myers Squibb, Celldex, CytomX, Genentech, Merck, Neon Therapeutics, Novartis, and Pfizer; and a speaking engagement from Medscape. F. Stephen Hodi reports the following: grants from Bristol-Myers Squibb and Novartis; personal fees from Bristol-Myers Squibb, Merck, Serono, Novartis, Takeda, Surface Pharmaceuticals, Genentech/Roche, Compass Therapeutics, Apricity, Bayer, Aduro, Partners Therapeutics, Sanofi, Pfizer, Pionyr Immunotherapeutics, 7 Hills Pharma, Verastem Oncology, Rheos Medicines, and Kairos Therapeutics; equity in Torque Therapeutics; and issued (#20100111973 and #7250291) and pending patents (#20170248603, #20160340407, #20160046716, #20140004112, #20170022275, #20170008962, and “Methods of Using Pembrolizumab and Trebananib”). Osama E. Rahma reports research support from Merck; educational grants from Bristol-Myers Squibb and Merck; consulting agreements with Bayer, Merck, Celgene, Five Prime Therapeutics, GlaxoSmithKline, GFK, Defined Health, Roche/Genentech, PureTech, Imvax, Leerink, Sobi, and PRMA Consulting; and a pending patent (“Methods of Using Pembrolizumab and Trebananib”). The other authors made no disclosures.

Author Contributions

Shilpa Grover: Conceptualization, data curation, methodology, writing–original draft, and writing–review and editing. Michael Dougan: Conceptualization, data curation, methodology, writing–original draft, and writing–review and editing. Kevin Tyan: Data curation, investigation, writing–original draft, and writing–review and editing. Anita Giobbie-Hurder: Data curation, formal analysis, methodology, software, and writing–review and editing. Steven M. Blum: Data curation, investigation, funding acquisition, and writing–review and editing. Jeffrey Ishizuka: Data curation, investigation, and writing–review and editing. Taha Qazi: Data curation, investigation, and writing–review and editing. Rawad Elias: Data curation, investigation, and writing–review and editing. Kruti B. Vora: Data curation, investigation, and writing–review and editing. Alex B. Ruan: Data curation, investigation, and writing–review and editing. William Martin-Doyle: Data curation, investigation, and writing–review and editing. Michael Manos: Data curation, investigation, and writing–review and editing. Lauren Eastman: Data curation, investigation, and writing–review and editing. Meredith Davis: Data curation, investigation, and writing–review and editing. Maria Gargano: Data curation, investigation, and writing–review and editing. Rizwan Haq: Data curation, investigation, and writing–review and editing. Elizabeth I. Buchbinder: Data curation, investigation, and writing–review and editing. Ryan J. Sullivan: Formal analysis and writing–review and editing. Patrick A. Ott: Formal analysis and writing–review and editing. F. Stephen Hodi: Formal analysis and writing–review and editing. Osama E. Rahma: Conceptualization, data curation, methodology, writing–original draft, and writing–review and editing.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.