Ulcerative colitis during CTLA‐4Ig therapy in a patient with rheumatoid arthritis

The coexistence of rheumatoid arthritis (RA) and ulcerative colitis (UC) is exceptional, with only a few cases reported in the literature.¹,²,³,⁴ However, the current use of biological agents can modify the balance of inflammatory mediators and the lymphocytic profiles of the original disease, promoting the emergence of additional conditions. This seems to be the case for a patient with severe RA who developed UC while being treated with CTLA‐4Ig (abatacept).

We report the case of a 55 year old male with a seven year history of seropositive RA refractory to treatment with 20 mg/week methotrexate (MTX), 1.5 g/day sulphasalazine (SSZ), 400 mg/day hydroxychloroquine, and 100 mg/day azathioprine. The patient was included in an CTLA‐4Ig (10 mg/kg intravenously monthly) plus MTX clinical trial. All other disease modifying antirheumatic drugs were discontinued. During treatment, major clinical and serological responses were achieved (table 1). Fifteen months later, the patient developed diarrhoea, rectal bleeding, crampy abdominal pain, tenesmus, and weight loss. CTLA‐4Ig and MTX were discontinued. Stool search for bacteria, ova, and parasites yielded negative results; serological tests for human immunodeficiency virus, hepatitis B, and hepatitis C were non‐reactive. Colonoscopy revealed extensive ulcerative lesions in the sigmoid colon and rectum. Histology showed a dense lymphoplasmocytic inflammatory infiltration distorting the architecture of the mucosa; cryptitis with irregular basal branching and intraluminal abscesses were also apparent and Paneth cells were identified in crypt bases. A diagnosis of UC was made and oral mesalamine therapy was started at a dose of 1.6 g/day with a satisfactory response. The dose of mesalamine was gradually reduced over the next three months until discontinuation.

Table 1 Clinical and laboratory findings.

Measurement	CTLA‐4Ig therapy
	Baseline	12 months	15 months	19 months	26 months
Tender joint count	22/28	9/28	11/28	16/28	28/28
Swollen joint count	18/28	2/28	8/28	12/28	24/28
C reactive protein (mg/l)	22.7	7.3	5.6	42.7	50.2
DAS‐28	6.58	4.25	4.77	6.16	7.47
Antinuclear antibodies*	0	1:40	1:40	0	0
Anti‐dsDNA (UI/ml)†	42	13	107	45	30

*Antinuclear antibodies are expressed in dilutions. Indirect immunofluorescence on HEp‐2 slides.

†Anti‐dsDNA, anti‐double stranded DNA antibodies (reference range ⩽99 UI/ml).

Four months after CTLA4‐Ig withdrawal, the patient remained asymptomatic for UC and the search for blood in stool was negative. However, arthritis was reactivated and a step on regimen with 25 mg/week MTX, 3 g/day SSZ, 5 mg/day prednisone, and 20 mg/day leflunomide was initiated.

Diverse methods have been proposed to assess the likelihood of a causal connection between environmental exposure (including therapeutic drugs) and an undesirable adverse event.⁵ According to Miller and colleagues,⁶ a series of five primary attribution elements or pertinent questions should be addressed when suspected exposure may be related to a clinical outcome for a given patient: (1) temporal association; (2) lack of likely alternative explanations; (3) dechallenge (improvement in symptoms following discontinuation of the drug); (4) rechallenge; and (5) biological plausibility.

In our case, both temporal association and dechallenge suggest that CTLA‐4Ig could be related to the development of UC and autoantibody induction. Analysing other likely alternative explanations, the major infectious causes of bloody diarrhoea were excluded, and no evidence of vasculitis of the mesenteric arteries or amyloidosis was found. Rechallenge was not available because CTLA‐4Ig was not reinitiated.

Biological plausibility is also possible considering the following facts: CTLA‐4Ig was designed and has been used to block the interactions between CD80/CD86 on antigen presenting cells and CD28 on T lymphocytes, preventing their activation.⁷ However, this same mechanism could alter both the development and maintenance of regulatory T cells,⁸ which control intestinal inflammation. Also, absence of the cytoplasmic domain of CTLA‐4 and hence of the intracellular signal it mediates favours a Th2 biased phenotype.⁹ This could explain the emergence of UC and the presence of autoantibodies, both phenomena related to a Th2 response.¹⁰

Although four of Miller's primary attribution elements are fulfilled, we cannot prove causality in this single case, especially in the absence of rechallenge. Nevertheless, theoretically, CTLA‐4Ig could favour the emergence of Th2 mediated phenomena, an issue that needs to be elucidated.