The negative impact of opioids on cancer patients treated with immune checkpoint inhibitors: a systematic review and meta-analysis

As one of the most effective analgesics, opioids are essential for patients with cancer-related pain, even in the context of the opioid abuse crisis. The current meta-analysis aimed to identify whether concomitant exposure to opioids can affect the efficacy of ICIs and lead to a worse prognosis. PubMed, Embase, and the Cochrane Library were searched based on the PRISMA checklist, through April 2022, for the following terms: ("opioids" OR "concomitant medication") AND ("Neoplasm" OR "Carcinoma" OR "Cancer" OR "Tumor") AND ("Immunotherapy" OR "Immune Checkpoint Inhibitor" OR "PD-L1 Inhibitor" OR "PD-1 Inhibitor" OR "CTLA-4 Inhibitor"). The outcomes considered were overall survival (OS) and progression-free survival (PFS) calculated using the random-effects or fixed-effects model. After screening 531 studies, a total of 7 articles involving 2690 patients were eligible for quantitative analysis. The use of opioids was negatively correlated with OS (HR 1.75, 95%CI 1.32–2.31, P < 0.001; I2 = 81%, P < 0.001) and significantly reduced the PFS (HR 1.61, 95%CI 1.41–1.83, P < 0.001; I2 = 0%, P = 0.63) of patients treated with ICIs. Similar results were obtained in each subgroup analysis. While NSAIDs could lead to poor OS (HR 1.25, 95% CI 1.03–1.51, P = 0.02; I2 = 0%, P = 0.60) but not PFS (HR 1.11, 95% CI = 0.89–1.39, P = 0.36) for ICIs patients. And sensitivity analyses confirmed the reliability of the results. Opioids significantly reduced OS and PFS in patients receiving ICI therapy. Thus, the use of different types of opioids should be considered with caution, and it is necessary to actively develop alternative treatments.


Introduction
Since being approved in 2014, immune checkpoint inhibitors (ICIs), represented by PD-1, PD-L1, and CTLA-4, have significantly improved the prognosis of patients and changed the treatment paradigm for several solid tumors (Grivas et al. 2019;Hoos 2016;Ribas et al. 2018;Xiong et al. 2021). However, ICIs also show some shortcomings in clinical application, including inconsistent efficacy and susceptibility to influence by other factors. Some concomitant medications, such as antibiotics or proton pump inhibitors (PPIs), have been indicated to reduce the survival benefit of ICIs (Derosa et al. 2022;Hopkins et al. 2022;Pinato et al. 2019). Thus, the combined utilization of ICIs and other drugs deserves more attention.
Pain is a common symptom in advanced and metastatic cancer. Nearly 80% of patients who died of cancer suffered moderate-severe pain for an average of 90 days in lessdeveloped countries (Knaul et al. 2018;van den Beukenvan Everdingen et al. 2016). The World Health Organization (WHO) guidelines suggest the use of opioids in adults or adolescents with cancer-related pain based on clinical assessment and pain severity (WHO Guidelines Approved by the Guidelines Review Committee 2018). Thus, even in the crisis of opioid abuse, pain management using opioids is still necessary for cancer patients and remains the best clinical analgesic for cancer-related pain.
With the application of ICIs in the first-line therapeutic regimen, the use of opioids combined with ICIs became common, and there have been inconsistent findings in some studies (Botticelli et al. 2021; Iglesias-Santamaría 2020; Taniguchi et al. 2020). Therefore, it is urgently necessary to evaluate the actual effect of opioids on ICI efficacy. We wish to provide a reference for the pain management of cancer patients, especially for those who are treated with ICIs.

Literature search
A systematic search was conducted using the PubMed, Embase, and Cochrane Library databases as well as related references retrieved up to April 2022 using the following search terms: ("opioids" OR "concomitant medication") AND ("Neoplasm" OR "Carcinoma" OR "Cancer" OR "Tumor") AND ("Immunotherapy" OR "Immune Checkpoint Inhibitor" OR "PD-L1 Inhibitor" OR "PD-1 Inhibitor" OR "CTLA-4 Inhibitor"). The search strategy was shown in Supplementary Table 1.

Eligibility criteria
Following the PRISMA guidelines (Moher et al. 2009), the authors searched the database according to the retrieval strategy and independently evaluated all articles. The title and abstract of the search results were browsed, and the full text was read to determine whether it met the inclusion criteria. Discrepancies were resolved by consensus.
The inclusion criteria used for article selection were as follows: (1) adult patients with cancer receiving ICI treatment, (2) patients were treated with opioids before, during, or after ICI administration, and the control group was not treated with opioids within the corresponding period, and (3) the outcomes were the efficacy of ICIs, including overall survival (OS) or progression-free survival (PFS). The exclusion criteria were as follows: (1) conference abstracts, review papers, papers without original data, and studies with duplicate data; (2) studies published in languages other than English; and (3) full-text article was not available.

Data extraction
We extracted the following data from eligible studies: first author, publication year, country, study type, cancer type, sample size, patient characteristics, ICI type, opioid type, and outcomes. We extracted the hazard ratio (HR) and 95% confidence interval (CI) of the multivariate analysis in the included articles. For the studies that only provided a survival curve we referred to the Engauge Digitizer method reported by Tierney to extract HR and 95% CI indirectly (Tierney et al. 2007).

Quality assessment
The quality of each study was evaluated using the Newcastle-Ottawa scale (NOS) (Wells et al. 2000). In the NOS system, studies with scores ≥ 6 are defined as high quality.

Statistical analysis and visualization tools
The effect of opioids on the survival of patients treated with ICIs was explored, and the results were reported as HRs and 95% CIs. OS was the primary outcome, and PFS was the secondary outcome. Heterogeneity was identified using the Q test, and we estimated and quantified it by I 2 values (Higgins et al. 2003). When I 2 was > 50% and/or P < 0.10, heterogeneity was considered statistically significant. A random-effects model or fixed-effects model was selected according to the heterogeneity results. Subgroup and sensitivity analyses were performed to determine the potential factors underlying the heterogeneity. Publication bias was evaluated by funnel plot, along with Begg's and Egger's tests. If publication bias existed, trim-and-fill analysis was used to assess it. All p values were two-sided, and the significance level was set at P < 0.05. Review STATA 15.1 and R 4.1.2 were used for statistical analysis and visualization.

Study selection
A total of 531 studies were retrieved from the initial broad search through April 2022. Based on the inclusion and exclusion criteria, 7 articles (Botticelli et al. 2021;Cortellini et al. 2020;Gaucher et al. 2021; Iglesias-Santamaría 2020; Kostine et al. 2021;Miura et al. 2021;Taniguchi et al. 2020) were eligible for quantitative analysis, with a total of 2690 patients (Fig. 1). There were 620 opioid-treated patients and 2070 opioid-free patients. The most common cancer types were non-small cell lung cancer (NSCLC), melanoma, and renal cell carcinoma (RCC). Finally, five studies (Botticelli et al. 2021;Cortellini et al. 2020; Iglesias-Santamaría 2020; Kostine et al. 2021;Taniguchi et al. 2020) provided both OS and PFS, and the other two (Gaucher et al. 2021;Miura et al. 2021) only reported OS. The baseline characteristics of the included studies are shown in Table 1.

Quality assessment
According to the NOS criteria, two reviewers independently evaluated the methodological quality of the included studies.
Overall, all studies were considered medium or high quality, which was indicated by scores of at least six ( Table 1).

Impact of opioids on ICIs (PFS)
Opioids significantly reduced the PFS of patients treated with ICIs (HR 1.61, 95% CI 1.41-1.83, P < 0.001) without heterogeneity (I 2 = 0.0%, P = 0.63), as shown in Fig. 2b. Subgroup analyses based on ICI type, sample size, and country obtained similar results, which indicated that opioid use was associated with poorer PFS (Table 2). Sensitivity analyses reported that the results were not dominated by any single study ( Supplementary Fig. 1b).

Risk of publication bias
The funnel chart ( Supplementary Fig. 2) and the results of Begg's test and Egger's test analysis (Table 2) suggested that there was no significant publication bias except for in the overall analysis of PFS (P Begg's = 0.027, P Egger's = 0.012). Trim-and-fill analysis showed that publication bias did not affect the PFS results (HR 1.55, 95% CI 1.38-1.74, P < 0.001).    Fig. 3). Several studies have corroborated that some medications can directly or indirectly influence immunity or immunotherapy, (Angrish et al. 2021;Gandhi et al. 2020;Gaucher et al. 2021;Kostine et al. 2021;Miura et al. 2021;Svaton et al. 2020), which has attracted considerable attention. In this meta-analysis, we focused on the impact of opioids on the survival outcomes of ICIs in advanced cancer patients.
In recent decades, the opioid abuse crisis has led to severe financial and social burdens and has been one of the biggest challenges facing public health in the twenty-first century (Macintyre et al. 2022). Although prescription drug-monitoring programs (PDMPs) have reduced the prescription rate of opioids from 255 to 153 million in America, they have also limited the adequate use of opioids for patients with cancer-related pain. The current consensus is that pain management is essential for tumor patients, and opioids are preferred for moderate-severe cancer-related pain and can contribute to a high quality of life and adherence to therapy (Scarborough et al. 2018). Thus, it seems unethical to restrict or forbid the use of opioids for severe cancer-related pain, and some investigators suggest providing exemptions for opioids for patients with cancer. However, based on this article, we believe that prescription opioids should be used with caution for tumor patients treated with ICIs, which is a novel but crucial viewpoint that might improve the longterm survival of tumor patients.
This study was the first meta-analysis to systematically evaluate the clinical efficacy of opioids on ICIs and included seven articles published in the past three years. Our metaanalysis identified the adverse effects of opioids on the efficacy of ICIs, and the results showed that the use of opioids was negatively correlated with OS and PFS in cancer patients treated with ICIs. Considering the heterogeneity in cancer type, ICI type, sample size, and publication country, we divided the study into several subgroups for further analysis. Almost all subgroups consistently showed the negative effect of opioids on the prognosis of patients, except in the melanoma subgroup, which revealed that opioids had a worse trend without a significant difference in OS. The reason for this might be that only two studies were included in the analysis of this subgroup. Regrettably, we likewise found similar studies in two conference abstracts (Bironzo et al. 2019;Taniguchi et al. 2019), of whose results were consistent but were not included because of insufficient evidence.
Sensitivity analysis showed that two studies (Botticelli et al. 2021;Kostine et al. 2021) strongly influenced heterogeneity. In the study of Botticelli (Botticelli et al. 2021), ECOG-PS was an independent prognostic factor rather than opioid use, which reflects patients' health status and the ability to tolerate therapy (Vercellino et al. 2020). Considering that patients treated with opioids may be weaker and have more complications than others, there was significant collinearity between opioid use and ECOG-PS, which might be one of the main causes for the heterogeneity in our metaanalysis. In addition, opioids had various immunoregulatory levels according to different targets, and morphine and fentanyl were stronger than others (Lu et al. 2021;Maher et al. 2020;Mao et al. 2016;Zajączkowska et al. 2018). Two included articles (Kostine et al. 2021;Taniguchi et al. 2020) disclosed relevant details of the opioid types. Of these, the main opioid in the article by Taniguchi was oxycodone, with a utilization rate of 52.6%, but fentanyl and morphine had utilization rates of 18.4% and 15.8%, respectively (Taniguchi et al. 2020). Kostine's study (Kostine et al. 2021) referred only to morphine, which seemed to have more negative effects on ICIs than in other studies and acted as another source of heterogeneity in our meta-analysis.
Multimodal analgesia is a promising therapeutic strategy and is drawing increasing attention to the management of cancer-related pain (Avery et al. 2022). Based on previous studies, the opioids with weak or no immune modulation (buprenorphine, oxycodone, hydromorphone, and tramadol) should be considered for the combined utilization with morphine or fentanyl, which can reduce the immunosuppressive effect of opioids for ICIs patients (Lu et al. 2021;Maher et al. 2020;Mao et al. 2016;Zajączkowska et al. 2018). In addition, alternative drugs for chronic pain, including NSAIDs, antidepressants, and anticonvulsants, might be another choice (Che et al. 2021). A network meta-analysis reported that certain non-opioid analgesics and NSAIDs can serve as effectively as opioids for chronic cancer-related pain (Huang et al. 2019). In this study, we revealed that NSAIDs could lead to poor OS but not PFS for ICI patients. Even so, NSAIDs seem to have a better effect than opioids on OS. As one of the representative NSAIDs, aspirin was researched independently in some studies because of its anti-thrombogenesis. Thus, we also focus specifically on aspirin for its pain relief efficacy, and aspirin had no additional effect on ICIs in terms of either OS or PFS. Interestingly, a metaanalysis reported that acupuncture and/or acupressure was significantly associated with reduced cancer pain and could decrease use of analgesics, which deserves more attention (He et al. 2020).
There are several limitations to this study. First, our metaanalysis was based on retrospective studies. Considering the lower-level evidence and the number of included studies in some subgroups, the results should be interpreted with caution. Second, due to a lack of basic data, we could not perform an in-depth analysis in terms of opioid type, dosage, or drug exposure time, which might be the factors driving non-statistical heterogeneity. In addition, tumor staging is an independent factor for prognosis, but only two articles provided information on staging, which might affect the accuracy of the results (Cortellini et al. 2020;Iglesias-Santamaría 2020). Third, since one of the included studies did not perform a multivariate analysis, we used the method of Tierney et al. to extract the HR and 95% CI according to the survival curve, which might lead to a certain bias.

Conclusions
As one of the most effective analgesics, adequate application of opioids is essential for patients with cancer-related pain, even in the context of the opioid abuse crisis. However, our study showed that opioids were associated with poor prognosis in patients treated with ICIs. Thus, caution should be taken when prescribing a drug combination. It is necessary to clarify appropriate opioids based on immunoregulatory levels of ICI therapy and actively develop alternative drugs in future.
Author contributions Concept and design: L, G, Z. Acquisition, analysis, or interpretation of data: J, G, L. Drafting of the manuscript: J, G, Z, W. Critical revision of the manuscript for important intellectual content: L, Zhu. Statistical analysis: J, G, Z. Administrative, technical, or material support: L, L. Supervision: L, Z. All authors had full access to the data and take responsibility for the integrity of the data and accuracy of the analysis. K is the guarantor.
Funding This work was supported by Liaoning Revitalization Talents Program (Grant No XLYC1905004), and Shenyang High-level Innovation Talents Plan (Grant No RC190403) for Kai Li, and Liaoning