Patterns of steroid use in diarrhoea and/or colitis (D/C) from immune checkpoint inhibitors (ICPI)

Background

D/C is reported in up to 30% of patients treated with ICPI. Treatment algorithms advocate corticosteroids (CS) for moderate to severe symptoms. Patterns of endoscopic change and duration of CS are not described. Rates of CS toxicity in this group of patients are unclear.

Methods

Medical records of melanoma patients treated with ICPI at the Royal Marsden Hospital from 2010-2015 were reviewed. The grade, duration of CS and infliximab (INF) use was recorded for each D/C episode. Patients who had flexible sigmoidoscopy (FS) were labeled as having macroscopic + /-microscopic (macro), microscopic (micro) changes alone or no changes (normal). CS toxicities were noted.

Results

414 ICPI treatment episodes were undertaken in 353 patients. The rate of all-grade D/C was 23% (96/414): 27% (77/282) with ipilimumab, 8% (8/101) with anti-PD-1 agents and 38% (8/21) with combination ipilimumab + nivolumab. Median age 61 years, 54% were male.

Steroid use by Flexible Sigmoidoscopy Findings

	N	Received CS (%)	No CS	Median CS duration days (range)	UNK re CS duration	On CS at 3mths (%)	UNK re CS at 3mths (%)	INF (%)	Median days CS to INF (range)
Macro	27	24 (89)	3	69 (5-278)	4	10 (42)	4 (15)	8 (33)	15 (6-60)
Micro	8	7 (88)	1	51 (12-162)	0	2 (29)	0	0	na
Normal FS	9	8 (89)	1	40 (7-156)	0	1 (13)	0	2 (25)	11 (2-20)
Unknown	6	5 (8)	1	67 (36-91)	1	1 (20)	0	1 (20)	9

57% (55/96) required CS. 52% (50/96) underwent FS: 88% (44/50) received CS. Median duration of CS was higher in macro patients compared to a normal FS (69 vs 40 days; p = 0.07) and in micro patients compared to a normal FS (51 vs 40 days; p = 0.32; Table 1). 11% (11/96) of D/C was treated with INF and median CS duration in this population was 91 days, versus 52 days for those who received CS alone. Median time from first steroid to INF was 15 days (range 2-60). 25% (14/55) were on steroids at 3 months. At least 15% (8/55) developed new or worse diabetes, 13% (7/55) had mood change and 16% (9/55) developed an antibiotic-requiring infection.

Conclusions

We describe different phenotypes of D/C associated with ICPI therapy. Appearances at FS may predict for duration of CS use. All patients should be monitored for stigmata of CS use and considered for PJP prophylaxis. CS sparing strategies should be prospectively evaluated.

Clinical trial identification

not applicable

Legal entity responsible for the study

James Larkin

Funding

N/A

Disclosure

All authors have declared no conflicts of interest.