Risk of immune-related diarrhea with PD-1/PD-L1 inhibitors in different cancer types and treatment regimens

Objective: To compare the incidence and severity of diarrhea among different tumor types and treatment regimens, and also compared with CTLA-4 inhibitors in randomized controlled trials. Methods: MEDLINE, PMC database and EMBASE were retrieved until December 2018. Studies were eligible if they were randomized controlled trials and included participants undergoing PD-1/PD-L1 inhibitors for cancer, measured a treatment side effect of diarrhea, and reported quantitative data. The risks of diarrhea in PD-1/PD-L1 inhibitors were compared among different treatment regimens. Results: Totally 21 studies involving 11554 patients were included for meta-analysis. For all-grade diarrhea, the risk after the PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor combination was 1.90 times significantly higher than that of monotherapy, and the risk was 0.69 and 0.60 times significantly lower than that of monotherapy compared with chemotherapy and ipilimumab. The incidence of diarrhea was not significantly different between PD-1/PD-L1 inhibitor monotherapy versus placebo or between low-doses versus high-doses. For high-grade (grade ≥ 3) diarrhea, the risk after the PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor combination was 3.29 times significantly higher than that of monotherapy, the risk in PD-1/PD-L1 inhibitors monotherapy was 0.50 and 0.38 times significantly lower than chemotherapy and ipilimumab respectively. No significant difference was found in the incidence of diarrhea between PD-1/PD-L1 inhibitor monotherapy versus placebo or between low-doses versus high-doses whether in all-grade or high-grade group. Conclusions: PD-1/PD-L1 inhibitors have a lower risk of developing diarrhea than chemotherapy and CTLA-4 inhibitor. There is no direct relationship between the dose of PD-1/PD-L1 inhibitors and the risk of developing diarrhea.

Diarrhea is a common side effect of cancer treatment that, in severe cases, can lead to death or to patients having to stop lifesaving treatment because often there are no effective therapies to control the diarrhea. Diarrhea in cancer patients can quickly lead to life-threatening consequences such as dehydration, electrolyte imbalance, shock, etc. Compared to chemotherapy-related diarrhea the immunological preparation of PD-1/PD-L1 is prone to cause autoimmune digestive diseases such as ulcerative colitis, and may also cause side effects of diarrhea.
Given the clinical efficacy evidence for a wide spectrum of tumor types, the PD-1 ICI therapy is expected to be increasingly used by oncologists as a monotherapy or in combination with other drugs. Therefore, physicians in cancer immunotherapy must be familiar with the pathogenesis of diarrhea in different tumors and different treatment regimens, and provide useful information to optimize the management of this toxicity. At present, there is no complete description about the clinical experience of anti-PD-1/PD-L1-associated diarrhea patients, or about the management and outcome of this toxicity. Therefore, we conducted a meta-analysis of PD-1 inhibitors in cancer patients and compared the incidence and severity of diarrhea among different tumor types, different treatment regimens.

Literature selection and data extraction
Two researchers (Lei Zhao and Huihui Li) independently reviewed the databases Medline, PMC database and EMBASE to select potential relevant articles. Any discrepancy between them was resolved by consensus . The following medical subject heading  terms were used: PD-1, PDL1, CD274, programmed  death receptor 1, programmed death receptor ligand,  immune  checkpoint  inhibitor,  nivolumab,  BMS936558, pembrolizumab, MK-3475, MPDL3280A, atezolizumab, avelumab, MSB0010718C, durvalumab, and diarrhea. The databases were searched from the inception until December, 2018.
The inclusion criteria were: (a) phase I, II and III trials in cancer patients; (b) random assignment of participants to single PD-1/PD-L1 inhibitor treatment or other control therapy (e.g. ipilimumab, placebo); (c) reporting diarrhea events or event rate and sample size for any all-grade or high-grade (≥3) adverse events;(d) random controlled trial.
The following information was extracted by two independent reviewers (Lei Zhao and Huihui Li) from the included studies: first author, publication year, study name, clinical trial registration number, total number of patients, mean age, trial phase, treatment plan, tumor type, primary inclusion criteria, and numbers of patients with all grades and high-grade treatment-related diarrhea. The treatment regimens were classified as PD-1/PD-L1 inhibitor monotherapy, PD-1/PD-L1 inhibitor plus CTLA-4 inhibitor ipilimumab, chemotherapy, placebo, and ipilimumab. According to the different doses, monotherapy was divided into low-dose group and high-dose group.

Quality assessment
The two reviewers (Lei Zhao, Huihui Li) used the Jadad scoring method [41] to evaluate the quality of each included study from randomized (0 or 1), double-blind (0, 1 or 2), recorded loss of follow-up and/or exit (0 or 1) and assign hidden (0 or 1). A score ≥ 3 indicates high quality.

Statistical analysis
Meta-analysis for statistical analysis was performed using Stata12.1 (Stata Corp, College Station, TX, USA). Heterogeneity was analyzed by Q test, and I 2 <25%, 25%-75%, and >75% indicate mild, moderate and significant heterogeneity, respectively. In case of insignificant heterogeneity between studies indicated as P>0.05, a fixed effect model was used; otherwise a random effect model was used. The incidence of diarrhea was evaluated by relative risk (RR) and 95% confidence interval (CI), and the analysis results were represented by forest maps. Two-tailed p < 0.05 was considered significant. This meta-analysis has been registered on the PROSPERO website (Registration Number: CRD42018111834).

PD-1/PD-L1 inhibitor monotherapy versus placebo
Inclusion of the study and number of patient's high-grade diarrhea were consistent with previous all-grade diarrhea. As shown in Figure 3A, when compared with placebo, there was not a significant increase in the risk of high-grade diarrhea incidence for PD-1/PD-L1 inhibitor monotherapy (RR 0.85, 95%CI: 0.29-2.44, P=0.756).

PD-1/PD-L1 inhibitor monotherapy versus chemotherapy
Inclusion of the study and number of patient's high-grade diarrhea were consistent with previous all-grade diarrhea. Results showed a significant decreased in the risk of high-grade diarrhea after monotherapy (RR 0.50, 95%CI: 0.26-0.95, P = 0.035; Figure 3B). Of the 7 RCTs on non-small cell lung cancer, 5 were treated with nivolumab and 2 with pembrolizumab. The use of nivolumab or pembrolizumab seems to reduce the risk of diarrhea compared to chemotherapy, but the results are not significant (RR 0.58, 95%CI: 0.25-1.31, P = 0.190; RR 0.61, 95%CI: 0.03-13.01, P = 0.754, respectively).

PD-1/PD-L1 inhibitor monotherapy versus ipilimumab
Inclusion of the study and number of patients of high-grade diarrhea were consistent with previous grades of diarrhea. Results showed significantly decreased in the risk of high-grade diarrhea after PD-1/PD-L1 inhibitor monotherapy (RR 0.38, 95%CI: 0.18-0.79, P=0.009; Figure 3C). Our meta-analysis reveals that PD-1 antibodies (pembrolizumab or nivolumab) reduce the risk of severe diarrhea compared to ipilimumab.

Nivolumab plus ipilimumab compared to nivolumab monotherapy
Inclusion of the study and number of patients of high-grade diarrhea were consistent with previous grades of diarrhea. The tumor type was melanoma in all cases. Results showed no significant increase in the risk of high-grade diarrhea after nivolumab plus ipilimumab treatment (RR 3.29, 95%CI: 1.80-6.03, P=0.000; Figure 3E).

Study quality and publication bias
Fifteen trials were open label, whereas five trials were double blind controlled. The Jadad score ranged from 3 to 4. For RR of all-grade between PD-1/PD-L1 inhibitor monotherapy and chemotherapy or high-grade diarrhea between the monotherapy and ipilimumab, the Egger test suggested some evidence of publication bias. No evidence of bias was found in other comparisons of Egger tests (all P>0.05), or in all Begg tests (all P>0.05).

Discussion
Although an increasing number of clinical studies have confirmed the overall survival benefit of PD-1/PD-L1 inhibitors treatment, PD-1/PD-L1 inhibitors therapy increases the toxicity of the drug remains controversial, especially diarrhea. Our meta-analysis is the first large-scale analysis of different immunologic treatment regimens compared with chemotherapy or ipilimumab for the toxic side effects of diarrhea. In our research, the risk of all-grade diarrhea after the PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor combination was 1.90 times higher than that of PD-1/PD-L1 inhibitors monotherapy (P<0.05), and the risk was 0.72 and 0.60 times higher than that of chemotherapy and ipilimumab compared with monotherapy (P<0.05). Chemotherapy is the most prone to diarrhea, followed by PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor combination, and finally PD-1/PD-L1 inhibitors monotherapy. When compared with placebo, we did not observe a significant increase in the risk of all-grade or severe diarrhea incidence, and our meta-analysis also reveals that high-dose PD-1/PD-L1 inhibitor monotherapy did not increase the risk of all-grade or severe diarrhea when compared with low-dose(all P>0.05), which suggested that PD-1/PD-L1 inhibitor monotherapy is relatively safe. The risk of grade ≥ 3 diarrheas for PD-1/PD-L1 inhibitors alone significantly decreased than chemotherapy or ipilimumab, while the PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor combination significantly increase than monotherapy.
The basic principle of binding PD-1 / PD-L1 inhibitors and CTLA-4 inhibitors is that they have different mechanisms of action. Anti-CTLA-4 mainly acts on the lymph node area, restores the induction and proliferation of activated T cells, and resists PD-1 acts mainly on the periphery of the tumor site, preventing the tumor-infiltrating tumor-infiltrating PD-L1-expressing tumor and plasma-like dendritic cells from neutralizing cytotoxic T cells [53]. Our result found that patients taking CTLA-4 inhibitors ipilimumab had a significantly higher risk of developing diarrhea than those using PD-1 / PD-L1 inhibitors. This is likely to be related to the different mechanisms of action of the two drugs. Although the combination of PD-1/PD-L1 inhibitors plus CTLA-4 inhibitor has achieved good efficacy [25,50,52], the corresponding toxic side effects of combination therapy, especially diarrhea, are significantly higher than those of PD-1/PD-L1 inhibitors alone. Combination therapy with both CTLA-4 and PD-1 blockers raised the risk of GI toxicities to about 45% which is much higher than monotherapy [54]. The risk of diarrhea was significantly different compared to the use of PD-1/PD-L1 inhibitors alone and chemotherapy in different tumor types. In patients with NSCLC patients, the risk of diarrhea using PD-1 / PD-L1 inhibitors monotherapy is significantly lower than that of patients receiving chemotherapy. However, this result did not find in melanoma patients. Our results show that patients taking pembrolizumab or nivolumab have a slightly different risk of all-grade diarrhea compared with chemotherapy [11,12,47]. Furthermore, our study demonstrates no significant difference in the incidence of diarrhea between low-dose versus high-dose PD-1/PD-L1 inhibitors, which are consistent with another study [12]. This provides reliable evidence for further exploration of adjusting drug doses in future clinical trial design and clinical practice.
As far as we know, this is the systematic review including the largest number of RCTs for analysis of immune-related diarrhea. The present study has some limitations. Firstly, these relevant studies have applied PD-1/PD-L1 inhibitors to different treatment lines, and there may be inconsistencies in the underlying characteristics of the patients. Secondly, since the present study is based on a secondary analysis of the final results of each report, we were unable to obtain patient-level disease characteristics and variables, or to determine the specific risk factors associated with the development of immune-related diarrhea. Thirdly, our results were influenced by the limitations of individual clinical trial design. Some of the clinical trials included in the meta-analysis were open label, which may lead to subjective bias. Finally, clinical RCTs included in the meta-analysis had strict inclusion and exclusion criteria. The patients selected in the study were with good PS, but in clinical practice, a large number of patients suffered impaired organ dysfunction and/or functional status and may have a higher incidence of actual toxicity. In the future, large-sample RCTs are needed to compare the incidence and severity of PD-1/PD-L1 inhibitor associated diarrhea among more tumor types and among more combination regimens.
Overall, PD-1/PD-L1 inhibitors have a lower risk of developing diarrhea than chemotherapy and CTLA-4 inhibitor. There is no direct relationship between the dose of PD-1/PD-L1 inhibitors and the risk of developing diarrhea. This study provides reliable evidence for further exploring the combination of PD-1/PD-L1 inhibitors with other drugs in clinical trial design and clinical practice.