Frequency of bowel perforation and impact of bowel rest in aggressive non‐Hodgkin lymphoma with gastrointestinal involvement

Non-Hodgkin lymphoma (NHL) involves the gastrointestinal (GI) tract in ~10–15% of patients at diagnosis (d'Amore, et al. 1994). A potential complication is perforation, which may result in peritonitis, septic shock, multi-organ failure, prolonged hospitalization, delayed chemotherapy and death. Some clinicians pre-emptively hospitalize patients following the first cycle of chemotherapy, employ gut rest and total parenteral nutrition (TPN) from initiation of therapy until neutrophil recovery. The putative rationale is to facilitate early recognition and reduce peritoneal contamination in the event of perforation. Despite substantial cost with prolonged admission and TPN, the benefit is unknown. To address this question, medical records and imaging reports were retrospectively reviewed at four sites, two in Australia (Sir Charles Gairdner Hospital, Perth, Western Australia and Princess Alexandra Hospital, Brisbane, Queensland) and one each in Canada (British Columbia Cancer Agency, Vancouver) and Denmark (Aalborg University Hospital, Aalborg). During the period studied, the Australian sites used routine bowel rest, while the Canadian and Danish sites did not. The objectives were to determine the incidence and timing of perforation among patients with lymphoma and GI involvement, and to determine the impact of bowel rest and TPN use during the first cycle of chemotherapy on frequency of perforation and outcomes following perforation. The secondary objectives were to examine the association between histological subtype, site of GI tract involvement, timing of perforation and duration of inpatient admission. Patients were included with first presentation of aggressive lymphoma treated with curative intent chemoimmunotherapy and GI involvement by either direct tissue biopsy or positron emission tomography-computerized tomography (PET-CT) or computerized tomography (CT) reports. A total of 414 patients were identified who met inclusion criteria: 202 (49%) were treated as outpatients and 212 (51%) were inpatients. Of the 212 inpatients, 94 (44%) received bowel rest and TPN and 118 (56%) observation only (Fig 1). The baseline characteristics of patients are summarized in Table SI The median age was 65 years (range 17–94); 63% were male. Diffuse large B cell lymphoma (DLBCL) was the most common histology; histological diagnoses are summarized in Figure S1 The primary site of GI involvement was gastric in 188 (45%), small bowel in 172 (42%), large bowel in 51 (12%) and rectal in 3 (1%) patients. PET-CT and CT imaging were used in 221 (53%) and 193 (47%) respectively. GI involvement was confirmed by direct tissue biopsy in 350 (85%) patients and imaging in 62 (15%). All patients with B cell lymphoma received a rituximab-containing chemotherapy regimen. Bowel perforation was observed in approximately 10% of patients overall. Of note, 6·8% occurred at initial presentation or prior to chemotherapy. Excluding these, the median time from commencing chemotherapy to perforation was 33 days (2–877). By multivariable analysis, poor performance status (subhazard ratio [sHR] 1·9; 95% confidence interval [CI] 1·0–3·5), involvement of small bowel (sHR 3·3; 95% CI 1·4–7·4), large bowel (sHR 3·1; 95% CI 1·1–8·9) and enteropathy-associated T-cell lymphoma (EATL) (sHR 3·4; 95% CI 1·5–7·6) were associated with greater perforation risk. Excluding patients with EATL (of whom 7/17 [41%] perforated, all at initial presentation) 8·5% of patients perforated, only 3% of which occurred after the commencement of chemotherapy. The data are consistent with Vaidya et al. (2013) who observed a perforation rate of 9% in all lymphoma subtypes. Involvement of the small bowel or large bowel was associated with a significantly increased rate of perforation, with similar findings in previous retrospective studies and case series(Ara, et al. 2007, Vaidya, et al. 2013) (Fig 2A). This may be related to the anatomical structure of the gastrointestinal lumen, which is significantly thicker in the stomach and rectum. In two prospective studies including 197 patients with primary and secondary gastric DLBCL treated with chemotherapy with >5 years follow-up, no perforations have been reported (Aviles, et al. 2004, Bonnet, et al. 2007). However, depth of tumour involvement in the bowel wall has not been shown to be linked with an increased rate of perforation in primary intestinal DLBCL.(Chuang, et al. 2008) Patients with EATL can be considered a special situation, where the risk of perforation is exceptionally high and the role of bowel rest and TPN should ideally be evaluated in prospective studies (Fig 2B). With a median follow-up of 2·9 years, excluding patients who perforated prior to commencement of chemotherapy, bowel rest was not associated with differences in rates of perforation (3·3% vs. 3·4%, P = 0·91, Fig 2C), peritonitis(P = 0·34), surgery(P = 0·86) or overall survival (P = 0·88, Fig 2D). Of the patients who perforated after the commencement of chemotherapy, no deaths were directly attributed to bowel perforation or its sequelae. In contrast, in the Mayo series, 12% (3 of 25 patients) died from complications of bowel perforation after receiving at least one cycle of chemotherapy(Vaidya, et al. 2013). This may reflect improvements in supportive care benefitting in our more recently treated cohort. Using Australian estimates for TPN, blood electrolyte monitoring and central venous access, the estimated cost of admitting 10 patients per year for bowel rest would amount to USD$446 848 per year, assuming the median inpatient stay of 22 days and TPN duration of 14 days. The usual limitations of retrospective, non-randomized design apply to this dataset. However, this concern can be partly addressed as bowel rest was systematically applied in two centres (but not in two others), somewhat ameliorating the potential for selection bias. Secondly, temporal relationship and the cause of bowel perforation and death were difficult to determine as many patients who perforated had multiple concurrent issues at time of admission. Thirdly, despite the overall sample size, the number of patients who experienced perforation was small, limiting our power to detect differences in outcomes. Acknowledging these limitations, a prospective randomized addressing these question is unlikely to be performed. In summary, bowel perforation complicates ~10% of aggressive NHL with GI involvement, 3% after chemotherapy. Bowel rest is not associated with a difference in the rate of perforation or outcomes post perforation. Patients with aggressive NHL with GI involvement should be educated to present for urgent assessment in the event of symptoms suggestive of perforation, as routine hospitalization, bowel rest and TPN do not improve patient outcomes. C.K.C performed the literature review, analysed the data and wrote the paper; E.T collected data, co-wrote and reviewed the paper; H.M and W.K collected data, co-wrote and reviewed the paper; J.B provided statistical advice, co-wrote and reviewed the paper; Z.N assisted with patient identification and cost calculations, co-wrote and reviewed the paper, G.H, P.M, D.J, D.R and G.C co-wrote and reviewed the paper; D.V and T.E.G. provided input with research design, co-wrote and reviewed the paper; C.Y.C designed the research, co-wrote and reviewed the paper. DV: Research funding from Roche to support the Lymphoid Cancer Database. Honoraria and advisory: Roche, Celgene, Janssen, Seattle Genetics, Merck. TEG: Travel funding: Roche and Takeda. CYC: Research funding: Celgene, Roche, Abbvie, Gilead; Speakers Bureau: Roche, Janssen-Cilag, Takeda, Gilead; Advisory Board member: Janssen-Cilag, Bristol Myers Squibb, Gilead; Travel expenses: Bristol Myers Squibb, Amgen. All remaining authors have declared no conflicts of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.