Oliver Oey,Yu-Yang Liu, Angela Felicia Sunjaya, Daniel Martin Simadibrata,Muhammad Adnan Khattak,Elin Gray

Abstract

Key Words: Melanoma; Gut microbiome; Microbiota; Immunotherapy; Biomarker; Immune checkpoint blockade therapy

lNTRODUCTlON

Melanoma is the most lethal form of skin cancer accounting for 73% of skin-cancer related mortality and over 50000 deaths worldwide annually[1,2]. Survival for metastatic melanoma has significantly improved since the introduction of immunotherapy and targeted therapy with a 5-year survival rate of up to 50%[3-5]. Currently, the standard first-line therapy for metastatic melanoma include BRAFtargeted therapies and immune checkpoint blockade (ICB) consisting either anti-programmed death(PD)-1 monotherapy or combination of anti-PD-1 as well as anti-cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) therapy[6]. Despite the considerable benefit of ICB, 40%-60% of melanoma patients do not experience objective responses to the therapy[7-9]. Thus, tremendous efforts are now focused on identifying novel biomarkers which could accurately predict the subset of patients who would benefit from ICB[10-14]. These biomarkers include tumor mutational burden, cytokines, circulating tumor DNA, human leukocyte antigen, gut microbiota (GM) diversity and composition, among many others[15].

The GM is a community of 100 trillion microorganisms of more than 1000 species mainly bacteria but also, archaea, viruses and fungi which colonize the human intestines[16]. The relationship which exists between GM and the host is a mutualistic relationship where one benefits the other[16]. In return for the nutrients derived from the host, the GM performs numerous critical functions such as fermentation of dietary fiber into short-chain fatty acids; synthesis of vitamins; protection against pathologic gut microbes; and induction and regulation of the immune system[17,18]. The gut microbial balance is pivotal in the optimal functioning of all of these roles and thus any discrepancy in this delicate equilibrium could produce a state of dysbiosis which has been associated with many pathologies including cancer[19]. In the context of cancer, preclinical studies have demonstrated that some GM subpopulations have pro-tumorigenic effects, whereas others have tumor-suppressive effects[20-22].Additionally, the GM has also been shown to modulate response to chemotherapy and immunotherapy[23-25]. This could be linked to the role of GM in metabolizing anti-cancer compounds and regulating the host’s immune response[16]. Thus, GM has been studied intensely as a potential biomarker of response to ICB[12,26-31]. This is particularly relevant for melanoma, where ICB has become standard of care given its demonstrated pronounced effectiveness.

Studies investigating GM composition and/or diversity in patients with melanoma have identified distinct GM composition in responders to ICB compared to non-responders, offering hope of a novel biomarker for predicting response to ICB[12,26-32]. Additionally, studies exploring whether certain GM composition and diversity could be predictive of ICB-related colitis - one of the major factors of ICB treatment cessation and thus failure to derive full benefit of ICB - have also been conducted[27,33]. This systematic review will be the first to compile the existing data regarding the role of GM composition and diversity in predicting response to ICB and ICB-related colitis specifically in patients with melanoma. Notably previous reviews have combined multiple cancers.

MATERlALS AND METHODS

Literature search strategies

This review was conducted following the preferred reporting Items for systematic reviews and metaanalyses guidelines[34]. The review protocol was submitted to the international prospective register of systemic reviews (PROSPERO Registration number: CRD42021228018).

In this comprehensive literature search, original studies exploring the variation in GM community in fecal samples of melanoma patients who responded and did not respond to immunotherapy,experienced colitis and did not experience colitis were identified. Medline and Embase were searched for eligible papers published prior to December 2021 using the following search terms: (fecal OR gut)AND (microbiota OR microbiome) AND (melanoma) AND (immunotherapy OR checkpoint OR nivolumab OR ipilimumab OR pembrolizumab). OpenGrey and the Grey Literature Report were also searched for eligible unpublished papers and grey literature. The following keywords and its synonyms will be used for our search strategy: “fecal microbiota”, “melanoma”, “immunotherapy”.

Duplicate and irrelevant publication types such as symposium agendas were removed from the initial search results. Titles and abstracts of relevant publications were screened independently by Oliver O and Simadibrata DM based on inclusion and exclusion criteria stated below. Subsequently, reference lists within each relevant publication were examined for further pertinent studies. The full texts of these publications were then reviewed.

Inclusion criteria

Inclusion criteria for the systematic review included randomized controlled trials (RCTs), original cohort, case-control studies published in a peer-reviewed journal exploring GM diversity and composition in fecal samples from melanoma patients treated with ICB which can be anti-PD-1 and/or anti-PD-L1 and/or anti-CTLA-4. Studies included should assess treatment outcome and/or ICB-related colitis incidence following treatment with ICB. Treatment outcomes should be determined by RECIST criteria and/or progression free survival (PFS) and/or overall survival (OS) and ICB-related colitis confirmed by colonoscopy.

Only studies which utilized fecal samples obtained from human subjects receiving ICB were included. Studies which assessed treatment response to immunotherapy in animal models were excluded. Two reviewers (Oliver O, Liu YY) independently screened and read the full text of the included articles for eligibility.

Data extraction

Two investigators (Oliver O, Liu YY) independently reviewed the eligible studies and extracted data from each study. Extracted variables included title, first author, year of publication, number of participants, type of immunotherapy received, GM analysis method, and study outcomes (GM composition and diversity in responders/non-responders and ICB-related colitis/non-ICB-related colitis patients). Extracted GM composition data included a list of the GM at the level of phyla, class,order, family, genus and species, whereas extracted GM diversity extracted included alpha diversity or the Shannon index. Any discrepancies found by the investigators on data extraction were resolved by consensus. In addition, we conducted a relevant search by Reference Citation Analysis (RCA) (https://www.referencecitationanalysis.com).

Quality assessment

Non-randomized studies, including cohort studies, case-control studies and single-arm clinical trial that were included in this systematic review were independently evaluated by Oliver O and Simadibrata DM for any risk of bias using the Risk of Bias in Non-randomized Studies of interventions (ROBINS-I)assessment tool, a tool which assesses seven items: confounding, selection, intervention classification,deviation from intervention, missing data, measurement of outcome and selection of reported result.Each item was assessed according to the ROBINS-I guideline, where each bias domain can be classified as either low, moderate, serious or critical risk of bias, or no information mentioned.

RESULTS

Study selection and risk of bias assessment

The initial search from Medline, Embase, OpenGrey and Grey Literature retrieved 300 studies. After deduplication, the studies were screened by reviewing their abstracts and 10 articles selected for full assessment (Figure 1). One study by Vétizouet al[35] was excluded because while the fecal samples were obtained from patients treated with anti-CTLA-4, treatment response to ICB was assessed in aninvivomice model of melanoma following fecal transplantation rather than humans.

From the nine included studies, two studies were phase I clinical trials, while the remainder were prospective observational studies[26,28]. Unfortunately, no RCTs were available to date. According to the ROBINS-I assessment tool, all but one study was shown to have moderate risk of bias (Table 1). The study by Matsonet al[29] had a serious risk of bias as there was a lack of clarity regarding the definition of intervention used.

GM composition and diversity in predicting immunotherapy response

Eight studies assessed the role of GM composition and/or diversity and response to ICB in melanoma patients (Table 2). Seven studies compared the GM between responders and non-responders to ICB, and two studies analyzed the GM in patients undergoing fecal microbiota transplant (FMT).

The study by Chaputet al[27] assessing fecal GM composition of 26 metastatic melanoma patients prior to and post commencing anti-CTLA-4 therapy revealed that GM composition varied according to response. Patients showing long term response to therapy (nine out of 26 patients) were found with fecal samples with significantly higherFaecalibacteriumpercentages (P= 0.0092) while patients with poor clinical benefit had higher proportions ofBacteroides(P= 0.034). When patients were grouped based on their microbiota composition, those with high prevalence ofFaecalibacteriumand otherFirmicuteshad a longer PFS (P= 0.0039) and to a lesser extent longer OS (P= 0.051) relative to patients whose fecal samples were abundant withBacteroides. Additionally, these patient groups were noted to derive longterm clinical benefit compared to the latter (67%vs0%;P= 0.0017)[28].

In an analysis of stool samples from 42 metastatic melanoma patients prior to treatment with anti-PD-1 (n= 38) and anti-CTLA-4 (n= 4) therapy, Matsonet al[29] showed a significant difference in GM composition between responders (16 patients) and non-responders (26 patients) (P ＜0.01). In responders, eight microbial species namely,Enterococcus faecium,Collinsella aerofaciens,Bifidobacterium adolescentis,Klebsiella pneumoniae,Veillonella parvula,Parabacteroides merdae,Lactobacillussp. andBifidobacterium longumwere found to be more abundant in responders than in non-responders[29]. In nonresponders, two microbial species, specifically,Ruminococcus obeumandRoseburia intestinaliswere more abundant[29]. To further assess the applicability of GM composition as a biomarker of response to ICB,they explored the correlation between the ratio of total numbers of potentially “beneficial” and“nonbeneficial” operational taxonomic units (OTUs), and change in tumor size, as assessed by the RECIST[29]. Patients with an OTU ratio of greater than 1.5 demonstrated clinical response to ICB[29].

In another study by Gopalakrishnanet al[12], fecal samples of 43 metastatic melanoma patients prior to treatment with anti-PD-1 therapy were analyzed. In responders (30 patients), analysis of fecal samples revealed abundance of GM fromRuminococcaceaefamily of theClostridialesorder, whereas in non-responders (13 patients), abundance of GM from theBacteroidalesorder was noted[12]. Further analyses demonstrated thatFaecalibacteriumgenus was notably enriched in fecal samples from responders andBacteroides thetaiotaomicron,Escherichia coli, andAnaerotruncus colihominiswere enriched in non-responders[12]. In addition, to investigate durability of response, patients were stratified based on their fecal composition ofFaecalibacteriumgenus andBacteroidalesorder and correlated to their PFS[12]. Results demonstrated that patients withFaecalibacterium-enriched fecal samples have longer PFS than those with low abundance (P= 0.03) and patients withBacteroidales-enriched fecal samples have shorter PFS than those with low abundance (P= 0.05). Beyond specific microbial taxa, GM diversity, as assessed by Simpson's reciprocal index, was higher in responders compared to non-responders (P ＜0.01)[12]. Moreover, high GM diversity was significantly associated with anti-PD-1 therapy response,when compared to patient groups of intermediate diversity [hazard ratio (HR) = 3.60, 95% confidence interval (CI): 1.02-12.74,P＜ 0.05) and low diversity (HR = 3.57, 95%CI: 1.02-12.52,P＜ 0.05). Other important predictors of therapy response include abundance ofFaecalibacterium(HR = 2.92, 95%CI: 1.08-7.89) andBacteroidales(HR = 0.39, 95%CI: 0.15-1.03) in the fecal microbiome[12].

Table 1 Risk of bias assessment with Risk of Bias ln Non-randomised Studies - of lnterventions

Figure 1 Prisma flow diagram of study selection.

Peterset al[30] examined the correlation between GM taxa and PFS in pre-treatment fecal samples of 27 metastatic melanoma patients receiving anti-PD-L1 and/or anti-CTLA-4. GM which was associated with shorter PFS included generaBacteroidesandBilophila, and speciesBacteroides ovatus,Blautia producta, andRuminococcus gnavus, whereas those which correlated with longer PFS included generaFaecalibacteriumandParabacteroidesand speciesFaecalibacterium prausnitzii[12]. With regards to GM richness the authors compared the β-diversity or between-sample microbiome diversity relative to survival. Multivariate analysis adjusting for age, sex, BMI, stage, number of sites of metastases, and antibiotic use in the last 6 mo revealed that higher GM richness was correlated with longer PFS (number of 16S sub - OTUs: HR [95%CI] = 0.97 [0.95, 1.00],P= 0.02; number of shotgun subspecies: HR [95%CI] =0.89 [0.79, 0.99],P= 0.03)[30]. Furthermore, analysis of the 16S but not shotgun dataset showed that higher diversity of GM, as assessed by the Shannon index, was associated with longer PFS (P= 0.02)[12].

Table 2 Characteristics of included studies exploring link between gut microbiome composition and diversity and response to immunecheckpoint blockade therapy in metastatic melanoma patients treated with immune-checkpoint blockade therapy

Similarly, Windet al[31] analyzed fecal samples from 25 metastatic melanoma patients - 12 responders, 13 non-responders - prior to start of treatment with anti-PD-1 or anti-CTLA-4. Analysis revealed that the fecal samples of responders were mainly enriched inRuminococcus gnavus,Escherichia coli,Eubacterium biforme,Phascolarctobacterium succinatutensandStreptococcus salivarius, whereas samples from non-responders were abundant inBifidobacterium longum,Prevotella copri,Coprococussp,Eggerthella unclassifiedandEubacterium ramulus[31]. When correlated with survival, fecal samples of participants enriched inBacteroides massiliensisandStreptococcus parasanguiniswere associated with longer PFS (HR:3.79, 95%CI: 1.06-13.52P= 0.04) and OS (HR: 5.05, 95%CI: 1.33-19.21,P= 0.017) respectively, whereas those who were carriers ofPeptostreptococcaceaewere associated with shorter PFS (HR: 0.18, 95%CI: 0.05-0.62,P= 0.007) and OS (HR: 0.12, 95%CI: 0.01-0.96,P= 0.046)[31]. In terms of GM diversity, as assessed by Shannon index, no significant difference between responders and non-responders was noted[31].

The study by Andrewset al[36] analyzed gut microbiome samples from a subset of 77 metastatic melanoma patients - 27 responders, 11 non-responders - who underwent combined ICB. There was no significant association inFirmicutesphyla andClostridialesorder and response to ICB (P= 0.39 andP=0.38, respectively) and no significant difference in alpha diversity between responders and nonresponders to ICB[36]. Fecal samples from responders were mainly enriched withBacteroides stercoris,Parabacteroides distasonisandFournierella massiliensis(P= 0.03,P= 0.04 andP= 0.008, respectively) while fecal samples from non-responders were abundant inKlebsiella aerogenes and Lactobacillus rogosae(P=0.04 andP= 0.02, respectively)[36].

In a first clinical trial of its kind (phase 1), Baruchet al[26] demonstrated that FMT from anti-PD-1 treated metastatic melanoma patients who were complete responders (2 donors), triggered response to anti-PD-1 therapy in metastatic melanoma patients who were refractory to at least one line of anti-PD-1 therapy. Out of 10 patients included in the trial, 3 patients demonstrated objective responses with 1 achieving complete response and 2 patients achieving partial response[26]. Notably, the PFS milestone of 6 mo was reached in all responders[26]. Upon analysis of pre-treatment fecal samples of donors,donor of the responding recipients had a lower microbial richness than the other donor of the nonresponding patients[26]. There was no significant difference on the GM composition prior to FMT of recipients who responded compared to those who did not respond[26]. Metagenome sequencing found that recipients post FMT have higher proportions ofVeillonellaceaefamily and a lower relative abundance ofBifidobacterium bifidum. Donors were found with high amounts ofLachnospiraceae, Veillonellaceae, andRuminococcaceae. Comparison of a small subset of non-responders with responders, found statistically significant higher abundance ofEnterococcaceae, Enterococcus, andStreptococcus australis, and a lower relative abundance ofVeillonella atypica. However clear deductions on specific GM taxa cannot be made, as there were non-responders and pre-treatment fecal samples with similar dynamics. it is crucial to note that this trial was primarily designed to assess safety of FMT and not statistically powered to assess efficacy[26].

In a separate trial, Davaret al[28] showed that fecal microbial transplant (FMT) from metastatic melanoma patients (7 donors) who had complete (4 donors) or partial response (3 donors) to anti-PD-1 therapy helped overcome resistance in anti-PD-1 treatment-refractory metastatic melanoma patients (15 patients). Following FMT and anti-PD-1 therapy, 6 out of 15 patients achieved clinical benefit, with 3 patients achieving objective responses and 3 patients experiencing stable disease lasting more than 12 mo[28]. Analysis of stools after FMT revealed that samples from responders were abundant in the phyla,Firmicutes(LachnospiraceaeandRuminococcaceae) andActinobacteria(BifidobacteriaceaeandCoriobacteriaceae) and had decreased proportions in phylumBacteroidetes[28]. In terms of GM diversity assessed with inverse Simpson index, GM diversity of donors who were complete responders were more diverse than donors who were partial responders. There was no significant difference in GM diversity between donors and recipients prior to FMT[28].

Gut microbiota composition and diversity in predicting ICB-related colitis

To date only three studies have reported on the correlation between pre-treatment GM composition and/or diversity and ICB-related colitis (Table 3).

Firstly, a prospective study by Dubinet al[33] explored the link between GM composition, and subsequent colitis development in 34 metastatic melanoma patients treated with ipilimumab, showed that theBacteroidetesphylum was more abundant (P= ＜ 0.05) in fecal samples of the 24 patients who did not develop ipilimumab-induced colitis compared to those who did. Further analysis revealed that within theBacteroidetesphylum, the population ofBacteroidaceae, RikenellaceaeandBarnesiellaceaewas significantly more abundant in the former than the latter (P ＜0.01,P ＜0.05 andP ＜0.05 respectively)[33]. However, there was no significant difference in microbial richness and diversity, as assessed by Shannon and inverse Simpson indices, between those who developed ipilimumab-induced colitis relative to those who did not[33].

In a similar study by Chaputet al[27], analysis of fecal samples of metastatic melanoma patients receiving ipilimumab demonstrated high proportions ofFirmicutesin patients who developed ipilimumab-induced colitis (P= 0.009). In contrast, fecal samples of those that did not develop colitis were mainly enriched withBacteroidetes(P= 0.011)[27]. Accordingly, patients with the former GM composition also tend to have a shorter colitis-free cumulative incidence compared with patients with the latter composition[27]. Several OTUs known to be predictive to colitis such asF. prausnitziiL2-6,butyrate producing bacteriumL2-21 andG. formicilisATCC 27749 were associated with longer OS[27].

Finally, Andrewset al[36], analyzed gut microbiome samples in metastatic melanoma patients undergoing combined ICB and their link to ileitis and colitis events. No significant difference in alpha diversity was observed between those that did and did not develop colitis[36]. Fecal samples of patients developing colitis were enriched inBacteroides intestinalisandIntestinibacter bartlettii(P= 0.009 andP=0.009, respectively) while those that did not were abundant inAnaerotignum lactatifermentansandDorea formicigenerans(P= 0.016 andP= 0.06, respectively)[36]. For bothB. intestinalisandD. formicigenerans,associations with their risk of colitis were still maintained after adjustment using a logistic regression model [OR = 4.54 (95%CI = 1.06-24.7) and OR = 0.35 (95%CI = 0.082-1.35), respectively][36].

DlSCUSSlON

Our review of current reports assessing the GM composition relative to response to ICB, indicated high variability in the results and limited concordance on the organisms identified (Figure 2). Amongst thefew commonalities, we found that fecal samples enriched in organisms from theFirmicutesphylum (LachnospiraceaeandRuminococcaceaefamily) especially theFaecalibacteriumgenus were associated with ICB responders in 4 of 9 studies[12,27,28,31], whileBacteroidetesphylum was found in higher proportions in non-responders in 2 of the studies[12,30]. However, other than these two findings, there was no clear correlation between specific GM composition and response to ICB.

Table 3 Characteristics studies exploring link between gut microbiome composition and diversity and immune-checkpoint blockade therapy-related colitis in metastatic melanoma patients treated with immune-checkpoint blockade therapy

In fact, our analysis mainly identified inconsistencies in the GM composition reported to be associated with response to ICB. For instance,Bifidobacterium longumwas found to be abundant in responders in the study by Windet al[31], but found to be enriched in non-responders in the study by Matsonet al[29]. Some species from theFirmicutesfamily were found in both responders and nonresponders such asRoseburia intestinalis[29]. Similarly, species from theBacteroidalesorder were found in both responders and non-responders, such asBacteroides massiliensis[31]. The overlap in GM composition in responders and non-responders may suggest that the functional capacity of the GM may be more important than individual GM family/order/species in determining response to ICB[30].

In contrast to individual species or taxas, GM diversity have been heralded to a marker of good health[37]. Here four studies - Gopalakrishnanet al[12], Peterset al[30], Windet al[31] and, Andrewset al[36] - assessed its potential to predict ICB responsiveness. The two first studies demonstrated that higher GM diversity in the responder group compared to non-responder arm[12,30]. However, the other two studies, Windet al[31] and Andrewset al[36], found no differences in GM diversity between both groups. Nevertheless, in other cancer types such as renal cell carcinoma and non-small cell lung cancer,greater GM diversity has also been associated with improved responses to anti-PD-1 therapy[37-39].

Study results showing associations with GM diversity were consistent with previous studies which showed that greater GM diversity is prevalent in healthy state across multiple diseases, plausibly suggesting that a greater GM diversity produces the optimal immune environment needed for normal physiological functioning[40-42]. One major reason is the promotion of a favorable immune phenotype,as evidenced by the positive correlation between Shannon diversity index and several CD8+ T cell and NK cell signatures, required to produce a robust anti-tumoral response[38].

Previous studies have demonstrated that GM fromFirmicutesfamily andBacteroidalesorder play a significant role in mediating the response to immunotherapy in melanoma patients[12,27,29]. For instance, abundance ofFirmicuteswas associated with increased frequencies of CD4+, CD8+ T cells, CD 45+ myeloid and lymphoid tumor-infiltrating cells and preserved cytokine response to anti-PD-1 therapy[12]. Additionally, abundance ofFirmicuteswas linked with decreased frequency of intestinal and systemic regulatory T cells (Tregs) and B7+T cells, cells responsible for limiting immune response robustness[27]. This resulted in increased antigen presentation and effector T cell function in both the periphery and tumor microenvironment[12,27,29]. However, other GM such asBacteroidaleswere unfavorable in terms of anti-tumoral response in that its abundance was associated with higher frequencies of Tregs and myeloid-derived suppressor cells and a blunted cytokine response[12]. These findings combined demonstrated that certain GM play a crucial role in mediating systemic and antitumor immune responses which have clear implications on efficacy on ICB therapy in metastatic melanoma patients.

Notably, GM has also been shown to potentially serve as not just a predictor of ICB therapy response,but also for boosting response to ICB therapy. FMT on anti-PD-1 treatment-refractory metastatic melanoma patients produced a complete response to anti-PD-1 therapy in one-third (9 out of 25 patients) of the otherwise therapy refractory patients[26,28].

Figure 2 Phylogenetic tree showing family and species of gut microbiome abundant in responders and non-responders to immunecheckpoint blockade therapy in all included studies. Gut microbiome species highlighted in red: Abundant in non-responders to immune-checkpoint blockade therapy; blue: Abundant in responders to immune-checkpoint blockade therapy; purple: Abundant in both responders and non-responders.

Another aspect of the GM analyzed here, was its association with ICB-related colitis. The three studies included in this review demonstrated that GM which was abundant in ICB-related colitis-prone patients was enriched in responders to ICB (Firmicutes) while GM which was abundant in ICB-related colitis-resistant patients was enriched in non-responders to ICB (Bacteroidetes). This is consistent with the understanding that a more effective anti-tumoral response will produce greater off-target effects. TheBacteroidetesphyla has been linked with low-grade systemic inflammation, which could explain the observation thatBacteroidetesphyla was abundant in ICB-related colitis-resistant patients[27]. In line with this observation is the finding that level ofBacteroidetesis lower in inflammatory bowel disease - an autoimmune condition which produces chronic inflammation of the digestive tract - patients relative to healthy patients[43]. Conversely,Firmicutesphyla, especiallyF. prausnitziihas been associated with induction of Tregs which express high levels of CTLA-4, fueling speculation that it may cause sequestration of Tregs within the intestine[44]. Since Tregs express high levels of CTLA-4, their actions are inhibited, thereby limiting self-tolerance and promoting the development of colitis. These findings reiterate that GM has an immunomodulatory role, giving them the potential to be utilized as biomarkers of ICB-related colitis, in addition to response to ICB.

Our systematic review has several strengths. Firstly, unlike previous reviews which combined studies in various cancer types, this review focused solely on the effect of GM composition and diversity only in patients with melanoma. Secondly, we conducted a comprehensive search for RCTs and observational studies, performed a risk-of-bas assessment and studied clinically important outcomes - clinical response and ICB-related colitis - an adverse event reported in up to 25% of patients treated with ICB[45]. Thirdly, we only included studies which assessed response to immunotherapy in humans, not animals. However, several limitations exist in our systematic review. Studies which we included used distinct approaches when segregating patients into the responder and non-responder groups, using different response criteria to evaluate treatment response in patients. Additionally, there were differences in methods of stool collection and analysis of GM composition and diversity. For example,Chaputet al[46] collected multiple stool samples every 3 wk of ICB, while other studies such as Dubinet al[33] and Matsonet al[29] collected stool samples only prior to initiation of ICB. Furthermore, only 4 studies considered confounding factors such as variation in diet and antibiotic use[27,29-31]. Therefore,inter-study comparison of the GM composition and diversity in respondersvsnon-responders and those who experienced colitisvsnon-colitis should be addressed with caution. Furthermore, included studies only enrolled a small number of patients, which could explain inconsistent results between studies.

CONCLUSlON

In conclusion, GM composition and diversity holds some potential as a biomarker of response and toxicity to ICB in melanoma. Larger prospective studies with standardized experimental protocol ought to be conducted to elucidate whether distinct GM signatures are required for robust response to different ICB regimens. Additionally, more studies correlating metagenomic and metatranscriptomic data of GM to outcomes of melanoma patients on immunotherapy ought to be performed as the functional capacity may be more important rather than individual GM family/order/species. In addition, we eagerly await the outcome of multiple large-scale RCTs involving FMT in the context of ICB-refractory melanoma such as NCT04577729 and NCT04988841 (PICASSO) (ClinicalTrails.gov).We foresee that together with other promising biomarkers, GM composition and diversity will be integrated into a multiparameter model to accurately predict which subset of melanoma patients are likely to respond to ICB[10,11,47].

ARTlCLE HlGHLlGHTS

Research conclusions

GM composition and diversity holds some potential as a biomarker of response and toxicity to ICB in melanoma. Further prospective studies, including several RCTs that are underway, are needed to confirm whether the GM could be used as a biomarker and potential intervention to modulate ICB response in melanoma patients.

Research perspectives

With other promising biomarkers, GM composition and diversity holds potential to be integrated into a multiparameter model to accurately predict which subset of melanoma patients are likely to respond to ICB.

FOOTNOTES

Author contributions:Oey O, Simadibrata DM, Gray E and Khattak MA contributed to the study conception and design; Oey O and Liu Y performed data extraction; Oey O and Simadibrata DM performed risk of bias assessment;Oey O, Liu Y, Sunjaya AF, Simadibrata DM, Khattak MA and Gray E performed data analysis; Oey O written the first draft of the manuscript; all authors commented on previous versions of the manuscript, read and approved the final manuscript.

Conflict-of-interest statement:Khattak MA reports receiving travel support from Merck Sharp and Dohme (MSD),Bristol-Myers Squibb and Merck Serono. Gray E reports receiving travel sponsorship from MSD. Oey O, Liu Y,Sunjaya AF, and Simadibrata DM report no competing interests.

PRlSMA 2009 Checklist statement:All authors have read the PRISMA 2009 Checklist, and the manuscript was prepared and revised according to the PRISMA 2009 Checklist.

Open-Access:This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BYNC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is noncommercial. See: https://creativecommons.org/Licenses/by-nc/4.0/

Country/Territory of origin:Australia

ORClD number:Oliver Oey 0000-0003-0673-6804; Yu-Yang Liu 0000-0002-7438-7258; Angela Felicia Sunjaya 0000-0001-8831-0449; Daniel Martin Simadibrata 0000-0002-7512-2112; Elin Gray 0000-0002-8613-3570.

Corresponding Author's Membership in Professional Societies:American Society of Clinical Oncology; American Association for Cancer Research.

S-Editor:Wang LL

L-Editor:A

P-Editor:Wang LL