Background
Gastrointestinal (GI) neoplasms threaten human health and account for approximately 35% of all cancer-related mortalities among common malignancies [
1]. Typically, patients are diagnosed accidentally with latent, unspecific symptoms reducing the already limited number of possible interventions. Surgical resection can be curative; however, the majority of patients are diagnosed in the advanced stages of this condition, therefore the opportunity for a radical cure is lost. The prevalence and impact of this insidious disease as well as limited treatment options necessitates the systematic search for innovative evidence-based treatments.
Advances in our understanding of immune-system/tumor interactions have led researchers to uncover new diagnostic pathways which may result in earlier identification. Also, several immunotherapies for the treatment of GI tumors have recently emerged. Among these new interventions, immune checkpoint inhibitor therapies are perhaps the most promising strategy [
2]. Indeed, the findings from many clinical trials suggest that immunological checkpoint blockade therapies may be effective for various types of tumor, with durable responses and manageable toxicity, regardless of pathologic grade [
3]. For those with GI tumors, blocking programmed cell death protein-1 (PD-1/CD279) or the ligand PD-L1 is also effective in approximately 20–40% patients. Due to such outcomes and with this moderate success, PD-1/PD-L1 blockades have been approved by the FDA for advanced colorectal, gastric, and liver cancers.
In contrast to other tumors such as lung cancer and breast cancer, GI tumors have mesenchymal traits which hinder the infiltration of immune cells thereby crippling the antitumor response [
4]. Likewise, the immunotherapeutic effects upon digestive tract tumors vary substantially which is perhaps due to different molecular and immunological characteristics. As such, several researchers have called for GI tumors to be reclassified based upon molecular type rather than around anatomical systems and histological features only [
5]. Despite this call for change, high mortality rates associated with these malignancies continues to drive clinical research in this field. Several phase I–III trials focusing on immunotherapies for GI tumors have found what can only be described as unsatisfactory objective response rates (ORR), ranging between 10 and 25% [
6]. In addition, problems such as drug resistance and the side effects of anti-PD-1/PD-L1 treatments remain challenging [
7]. So, while this growing body of evidence suggests that target-driven treatment strategies are essential, there is a paucity of research from which to design new interventions.
Presently, the logical next step appears to be combining immunotherapies with antitumor drugs and some progress has been made in preclinical and clinical studies which suggest that combined immunotherapies may increase benefit. However, this is a relatively new field of study hence effort should be made to embed research systematicity using secondary literature. As such, this study focuses on reviewing the current limitations of immune checkpoint blockade monotherapies and to critically discuss the rationale behind combination strategies based on the PD-1/PD-L1 blockade. The aim is to provide researchers and practitioners with a summary of the clinical applications of combination therapies for patients with upper and lower GI tumors and to explore the arguments around combination immunotherapies.
PD-1/PD-L1 pathway blockade: current limitations in clinical treatment
The immune checkpoint pathway composed of PD-1/CD279 and the related ligand PD-L1 evade immune surveillance by upregulating the expression in tumor cells during the progress of T cell-mediated immune killing. Substantial evidence from preclinical models indicates that blocking PD-1/PD-L1 interactions can enhance immune normalization and reinforce anticancer responses [
8,
9]. As early as 2003, Chen et al. found that using the B7 homolog 1 (B7-H1) blocking antibody combined with T cell transfusion cured approximately 60% of the 24 mice with squamous cell carcinomas in the head and neck. Without the transfusion of T cells, only one of five mice treated with B7-H1 blockade had prolonged survival; however, this was not considered a statistically significant improvement compared with the control group [
10].
In 2012, a phase I clinical trial investigating the efficacy of pembrolizumab for patients with advanced tumors found that the objective response rate (ORR) for patients with advanced non-small cell lung cancer (NSCLC), malignant melanoma, and advanced renal carcinoma was 18%, 28%, and 27%, respectively, and the adverse event profile does not appear to preclude its use [
11]. Similarly, a longitudinal study focusing on pretreated advanced NSCLC, involving 129 patients found a 16% five-year survival rate. While this study contained a larger number of participants which adds precision, pretreatments were not standardized. Nevertheless, this study suggests that PD-1 blockading may prolong therapeutic durability [
12]. This evidence of antitumor activation and the antibodies targeting the capabilities of PD-1/PD-L1 convinced the FDA to officially approve five inhibitors. The preliminary indications were that these inhibitors could be administered for several different types of tumor, including microsatellite instability-high (MSI-H) solid tumors.
The main advantages of PD-1/PD-L1 inhibitors are effect persistence (i.e., durability) and the broad-spectrum effects of these agents. However, the noticeable deficiency of PD-1/PD-L1 blockades is
inconsistency across a homogeneous study population with similar tumor characteristics [
13]. The exception to this can be observed in tumors with specific genetic changes, such as MSI-H, deficient mismatch repair (dMMR), and high tumor mutational burden (TMB). A review of the status and perspectives of translational biomarkers found the ORR is only 15–25% for unscreened solid tumors and even lower for some tumors, such as colorectal and pancreatic cancer [
14] which suggests the causal factor for this relatively low response rate might be attributed to tumor heterogeneity, genetic variation among individuals, and perhaps structural differences between blockades [
15]. Although, studies have also found that the development and evolution within a tumor itself can lead to a decreased efficacy of the PD-1 blockade. This may be due to genetic alterations within DNA encoding immunogenic signaling pathway proteins, a lack of sufficient mutation-associated neoantigens (MANAs) in the presence of an immunosuppressive tumor microenvironment, and/or the unmasking of immunogenicity by immune checkpoint inhibitors (ICPIs) to induce an enhanced antitumor response [
16].
As well as increasing antitumour activity, PD-1/PD-L1 blockade treatments may also cause certain inflammatory side effects in some patients which are referred to as immune-related adverse events (irAEs) [
17,
18]. Essentially, these immunotherapies unbalance the immune system, generating dysimmune toxicities which potentially effect any tissue. However, a systematic review of the side effects of the PD-1/PD-L1 blockade suggests irAEs can be widespread but are more likely to involve the GI tract, endocrine glands, and skin [
19‐
21]. Compared to the side effects of chemotherapy, immunotherapeutic side effects appear more diverse, random, and differential but primarily organ-based manifestations [
17]. Some studies indicate that these irAEs may be closely related to the expression and distribution of PD-L1 and PD-L2 [
22‐
24] which suggests while irAEs may be heterogeneous in nature, they may be tolerable and most associated side effects are treatable. However, there are potentially serious adverse reactions, such as myocarditis which can cause death. A substantial increase in the number of deaths associated with immune checkpoint inhibitors has been observed, although this may be attributed to increased use and raised awareness of this clinical entity [
25]. Conversely, some irAE studies have found improved immune responses in patients which suggest that these might also be used to predict treatment efficacy [
26].
The efficacy of PD-1/PD-L1 blockades can be lasting for some patients, although tumor development remains a constant threat even under continuous therapy [
27]. In a screening evaluation of PD-1 for the treatment of malignant melanoma, 48 cases were found to have significantly reduced tumor size or stable progression. However, in approximately half of those participants, tumors initially shrank before increasing in size directly after receiving this intervention [
28]. This suggests that this treatment may have had little or no effect overall due to immunotherapeutic resistance. At present, the possible mechanisms of acquired immunotherapy resistance appear to include loss-of-function mutations in beta-2-microglobulin (B2M) and Janus kinases (JAK1 and JAK2) [
29].
A study of two fully immunocompetent mouse models focusing on lung adenocarcinoma indicate that the T cell immunoglobulin mucin-3 (TIM-3) was upregulated in tumors resistant to PD-1 blockade, and a survival advantage was found with the addition of a TIM-3 blocking antibody following failure of the PD-1 blockade. This suggests that there may be a targetable biomarker associated with adaptive resistance to PD-1 blockades [
30]. Early clinical investigations have also found some patients with complete remission after treatment with PD-1/PD-L1 blockades, relapse. Although, data related to this phenomenon is limited, it does suggest a lack of therapeutic durability in humans which is supported by basic medical evidence.
Adding to the aforementioned side effects and drug resistance after immunotherapy, studies indicate that a small number of patients on PD-1 blockades will experience hyper-progression [
31‐
33]. The Ferrara study, which included 242 patients, found that tumor growth rates increased by more than 50% in 16% of patients (
n = 40) after receiving the PD-1 antibody. This finding meets the criteria for hyper-progression; however, this study lacked a control group and determining tumor progression causality was not possible [
34]. To explore this phenomenon in more detail, Singavi et al. conducted an analysis of somatic alterations looking into the biomarkers for hyper-progression and found that copy number alterations in murine double minute 2/4 (MDM2/MDM4), the epidermal growth factor receptor (EGFR), and several genes located on 11q13 are associated with hyper-progression. The role of these somatic alterations as putative predictive biomarkers for hyper-progression requires further investigation with larger samples [
35].
Identifying biomarkers is crucial as these might support both treatment efficacy and AE predictions in patients receiving immunological checkpoint therapy [
36]. Biomarkers such as dMMR and MSI, TMB and blood TMB, HLA diversity and PD-L1 expression have been explored. While stable predictors are not, presently available, different regions of the body develop different types of tumor, therefore antibodies used for detecting PD-L1 expression may be highly specific to one region while insensitive to level of expression, and vice versa. Furthermore, the activation effect of subsequent treatments is likely to change PD-L1 expression [
37], a factor which is currently adopted in clinical trials to predict immunological efficacy [
38]. While TMB, dMMR, and MSI positively correlate with the efficacy of PD-1, they are not widely used due to the limitations of these detection techniques [
39]. In summary, our knowledge of these biomarkers is far from complete therefore cannot be used as guidelines for precision immunotherapy. Alternative predictive markers are currently in the early exploratory phase [
40,
41].
Current applications of combination immunotherapy in gastrointestinal tumors
Among the cluster of digestive tract tumors, histological differences are significant and are generally used to determine which approach to implement, especially for advanced tumors. For example, radiotherapy is efficacious in patients with esophageal cancer but not in patients with pancreatic cancer. Likewise, chemotherapy is the main stay for the treatment of patients with advanced gastric cancer but chemotherapeutic regimens are not generally administered for hepatocellular carcinoma. Therefore, combining superior interventions for digestive tract tumors with a single-drug immunotherapy may achieve enhanced immune expansion, despite the efficacy of PD-1/PD-L1 blockades varying substantially. Most studies are in the early phase clinical trials, although there are some which have progressed to phase III (Table
1). In this section, we systematically review officially published clinical studies for GI cancer sought through
clinicaltrial.gov, PubMed, and in gray literature including conferences, such as ASCO and ESMO. Levels of efficacy will be critical discussed for several major digestive system tumors using relevant treatment indexes (i.e., OS, PFS, etc.).
Table 1
Ongoing phase 3 clinical trials of combined immunotherapy in gastrointestinal cancers
Unresectable, recurrent, locally advanced or metastatic gastric or gastroesophageal junction adenocarcinoma | 3/371 | PD-L1 inhibitors | Avelumab + BSC VS physician’s choice + BSC | Active, not recruiting | NCT02625623 |
Various advanced cancers | 3/939 | PD-1 and CTLA-4 inhibitors | Nivolumab + ipilimumab or nivolumab + fluorouracil + cisplatin VS fluorouracil + cisplatin | Recruiting | NCT03143153 |
Esophageal neoplasms | 3/700 | PD-1 inhibitors | Pembrolizumab + cisplatin and 5-fluorouracil (5-FU) VS placebo + cisplatin and 5-FU | Recruiting | NCT03189719 |
Esophageal carcinoma| esophagogastric junction carcinoma | 3/720 | PD-1 inhibitors | Pembrolizumab (MK-3475) VS Investigator’s Choice Standard Therapy | Active, not recruiting | NCT02564263 |
Gastric cancer | 3/700 | PD-1 inhibitors | Nivolumab + S-1 therapy or CapeOX therapy VS placebo+ S-1 therapy or CapeOX therapy | Recruiting | NCT03006705 |
Gastric cancer | 2,3/680 | PD-1 inhibitors | ONO-4538 + chemotherapy | Active, not recruiting | NCT02746796 |
Gastric cancer| gastroesophageal junction cancer | 3/1649 | PD-1 and CTLA-4 inhibitors | Nivolumab + ipilimumab or nivolumab + chemotherapy VS chemotherapy alone | Recruiting | NCT02872116 |
Gastric cancer| Gastroesophageal junction cancer | 3/860 | PD-1 inhibitors | Pembrolizumab (MK-3475) + chemotherapy VS placebo + chemotherapy | Recruiting | NCT03221426 |
Stomach neoplasms | 3/780 | PD-1 inhibitors | Pembrolizumab (MK-3475) + chemotherapy VS placebo + chemotherapy | Not yet recruiting | NCT03675737 |
Gastric neoplasms| gastroesophageal junction adenocarcinoma | 3/732 | PD-1 inhibitors | Pembrolizumab/Placebo + trastuzumab + Chemotherapy | Recruiting | NCT03615326 |
Gastric adenocarcinoma | 3/764 | PD-1 inhibitors | Pembrolizumab as monotherapy, or pembrolizumab + Cisplatin + 5-fluorouracil (5-FU) or capecitabine; placebo + cisplatin + 5-FU or capecitabine | Active, not recruiting | NCT02494583 |
Biliary tract neoplasms | 3/390 | PD-1 inhibitors | KN035 + gemcitabine + oxaliplatin VS gemcitabine + oxaliplatin | Recruiting | NCT03478488 |
Hepatocellular carcinoma | 3/330 | PD-1 inhibitors | Pembrolizumab (MK-3475) or placebo + Best supportive care | Recruiting | NCT03062358 |
Hepatocellular Carcinoma | 3/480 | PD-L1 inhibitors | Atezolizumab + bevacizumab VS sorafenib | Recruiting | NCT03434379 |
Hepatocellular carcinoma | 3/1200 | PD-L1 and CTLA-4 inhibitors | Durvalumab + tremelimumab | Recruiting | NCT03298451 |
Pancreatic cancer stage IV | 2/40 | PD-1 inhibitors | Nivolumab + cabiralizumab + gemcitabine VS gemcitabine | Not yet recruiting | NCT03697564 |
Colorectal cancer | 3/363 | PD-L1 inhibitors | Cobimetinib + atezolizumab and atezolizumab monotherapy VS regorafenib | Active, not recruiting | NCT02788279 |
Colorectal adenocarcinoma| mismatch repair deficiency | 3/347 | PD-L1 inhibitors | Atezolizumab, bevacizumab, Mfolfox6 VS bevacizumab, Mfolfox6 VS atezolizumab | Recruiting | NCT02997228 |
Colon Adenocarcinoma| DNA repair disorder | 3/700 | PD-L1 and CTLA-4 inhibitors | Combination chemotherapy with or without atezolizumab | Recruiting | NCT02912559 |
Colorectal cancer | 3/180 | PD-1 inhibitors | Nivolumab with standard of care therapy VS standard of care therapy for first-line treatment | Recruiting | NCT03414983 |
Esophageal carcinoma
Moderate progress has been made in the diagnosis and treatment of esophageal cancer; however, the 5-year survival rate for patients with advanced esophageal cancer remains less than 15%. A PD-1 blockade is mainly administered for patients with advanced esophageal cancer, including patients showing first-line drug resistant esophageal cancer, or localized progression and advanced metastasis. In the USA, pembrolizumab has been approved for the treatment of patients with chemotherapy-refractory PD-L1-positive gastroesophageal junction cancers on the basis of clinical activity observed in the KEYNOTE-059 trial. This study found that 95 patients, representing 42.4%, experienced a reduction in measurable tumor size with a corresponding 11.6% ORR [
95]. However, the KEYNOTE-180 study also found a 14% ORR for PD-1 blockages in esophageal squamous cell carcinoma patients compared with that of esophageal adenocarcinoma patients which was only 5%. This finding was lower than had been expected given the findings in the KEYNOTE-28 study where the ORRs of squamous cell carcinoma and adenocarcinoma were 29% and 40%, respectively [
96,
97]. A subsequent phase III study, KEYNOTE-181 (NCT02564263), is currently looking to evaluate the activity of pembrolizumab versus a standard therapy in patients with metastatic esophageal carcinoma which progressed after receiving a first-line intervention. Preliminary outcomes suggest pembrolizumab is superior to chemotherapy for OS in PD-L1 with combined positive score ≥ 10 patients. The reported 12-month OS rate was 43% as opposed to 20%, and drug-related AEs associated with pembrolizumab were fewer than in the group which received chemotherapy alone (64% versus 86%).
ICPIs in esophageal cancer encourage optimism and combined with an immunotherapy may bring further benefit for those suffering esophageal cancer. Several clinical trials investigating PD-1 combined with radiotherapy have already been conducted with esophageal cancer patients. The CheckMate-032 study focused on double immunotherapeutic interventions for esophageal cancer and found a 24% ORR for nivolumab administered at 1 mg/kg plus ipilimumab administered at 3 mg/kg, with a corresponding PFS at 12 months of 17% (Table
2). This finding was twice that of the group which received nivolumab alone. However, the treatment-related grade 3 and 4 AEs in the combination group was 47%, whereas with the single agent intervention resulted in only 17%. The authors concluded that treatment with this ipilimumab combination significantly increased the incidence of side effects [
98]. Finding an appropriate combination is clearly required, further necessitating the development of this evidence-base.
Table 2
Key trials of combination immunotherapy in esophagogastric cancers
CheckMate-032; NCT01928394 | Metastatic esophagogastric cancer | Ph-1/2 | 160 | Nivolumab 3 mg/kg; nivolumab 1 mg/kg + ipilimumab 3 mg/kg; nivolumab 3 mg/kg + ipilimumab 1 mg/kg | PD-1 + CTLA-4 | ORR: 12%, 24%, 8%; PFS rate at 12 months: 8%, 17%, 10%; OS rate at 12 months: 39%, 35%, 24% | Grade 3/4: 17%, 47%, 27% | |
KEYNOTE-059 cohort 2; NCT02335411 | Advanced gastric cancer | Ph-2 | 25 | Pembrolizumab + 5-fluorouracil + cisplatin | PD-1 + Chemo | ORR: 60%; mPFS: 6.6 months; mOS: 13.8 months | Grade 3/4: 76% | |
NCT02999295 | Advanced gastric adenocarcinoma | Ph-1/2 | 46 | Ramucirumab + nivolumab | PD-1 + Target | PR: 22%; DCR: 59% | Any grade: 87%; Grade 3/4: 13% | |
NCT02013154 | Advanced gastroesophageal cancer | Ph-1 | 13 | Dickkopf-1 + pembrolizumab | PD-1 + Molecules | PR: 1 pt/9 pts; SD: 5 pts/9 pts; DCR at 6 weeks: 75% | Grade ≥ 3: 4 pts./9 pts | |
Attraction-04; NCT02746796 | Unresectable advanced or recurrent G/GEJ cancer | Ph-2/3 | 40 | Nivolumab + oxaliplatin + S-1,or capecitabine; placebo + oxaliplatin + S-1, or capecitabine | PD-1 + Chemo | ORR: 68.4%; DCR: 86.8%; | Grade 3/4: 57.5% | |
NCT02689284 | Advanced HER2+ gastric adenocarcinoma | Ph-1/2 | 60 | Margetuximab + Pembrolizumab | PD-1 + Target | ORR: 21.6%; DCR: 55%; mOS: 15.6 months | Grade ≥ 3: 16.7% | |
NCT02864381 | Unresectable or recurrent gastric or gastroesophageal junction adenocarcinoma | Ph-2 | 144 | Andecaliximab 800 mg + nivolumab 3 mg/kg, or nivolumab 3 mg/kg alone | PD-1 + Molecules | ORR: ADX + nivo, 11.1%; nivo alone, 6.9%; mPFS: ADX + nivo, 1.8 months; nivo alone, 1.9 months; mOS: ADX + nivo, 7.2 months; nivo alone, 5.9 months | AEs leading to treatment discontinuation: ADX + NIVO, 1 pt.; nivo, 1 pt | |
NCT02954536 | HER2-positive metastatic esophagogastric adenocarcinoma | Ph-2 | 24 | Pembrolizumab + chemotherapy/trastuzumab | PD-1 + Chemo/Target | ORR: 83%; mPFS: 11.4 months | Gr 2 fatigue (35%), Gr 2/3 nausea (35%), Gr 2 diarrhea (26%), Gr2 AST/ALT elevation (16%), Gr2 neutropenia (16%) | |
NCT02689284 | Advanced HER2+ (IHC3+) gastric carcinoma | Ph-1/2 | 66 | Margetuximab 15 mg/kg + pembrolizumab 200 mg | PD-1 + Target | ORR: 41.4%; DCR: 72.4%; mPFS: 5.5 months | Grade ≥ 3: 18.2% | |
NivoRam study; NCT02999295 | Advanced gastric adenocarcinoma | Ph-1/2 | 46 | Nivolumab 3 mg/kg, Q2W + ramucirumab 8 mg/kg, Q2W | PD-1 + Target | ORR: 26.7%; PFS rate at 6 months: 37.4%; mPFS: 2.9 months; mOS: 9.0 months | Grade 3/4: hypertension, diarrhea, perforation at jejunum, hemorrhage, colitis, pancreatitis, liver dysfunction, cholangitis, hematoma, neutropenia and proteinuria | |
Based on current findings, a further phase III studies (NCT02872116) was designed to evaluate double immunotherapy as an early line therapy for esophagogastric cancers, and is presently under way. For the PD-1 and chemotherapy combination, the NCT03189719 trial is ongoing to evaluate the efficacy and safety of pembrolizumab plus cisplatin and 5-fluorouracil (5-FU) chemotherapy versus placebo plus cisplatin and 5-FU chemotherapy as a first-line treatment in participants with locally advanced or metastatic esophageal carcinoma. In fact, the majority of trials in this field are still in exploratory phases involving a variety of combinations. While results are pending, current knowledge provides some optimism and the results are eagerly anticipated.
Gastric carcinoma
The Cancer Genome Atlas (TCGA) divides gastric cancer into an Epstein-Barr virus (EBV) positive subtype, a microsatellite instability (MSI) subtype, a genomically stable (GS) subtype, and the chromosomal instability (CIN) subtype, according to histologically based integrative genomics [
108]. Among the four types of gastric cancer, the high-frequency MSI (MSI-H) subtype appears to respond favorably [
109]. The results of the ATTRACTION-02 phase III study focusing on heavily pretreated patients with advanced gastric or gastroesophageal junction cancer found OS rates in nivolumab compared with placebo were 27.3% and 11.6% at 12 months, and then 10.6% and 3.2% at 24 months, respectively. However, the nivolumab ORR was only in 11% of 268 patients which was considered a relatively low response rate [
110].
Comparatively, the KEYNOTE-061 trial which focused on pembrolizumab with paclitaxel in patients with advanced gastric cancer whom had developed resistance after platinum and fluoropyrimidine treatment found that pembrolizumab did not significantly improve OS compared to paclitaxel, with an 9.1 month mOS versus 8.3 months [
111]. Unsatisfactory immune monotherapies in gastric cancer make combined therapy especially enticing. Although, most of the combination strategies being investigated in gastric cancer are in the preclinical or early clinical research stage, few have entered the phase III stage [
112]. For example, the CheckMate-649 is further assessing the difference in survival between nivolumab plus ipilimumab and chemotherapy although results are pending.
In the KEYNOTE-059 cohort 2 study, the ORR and DCR of 25 patients with advanced gastric or gastroesophageal adenocarcinoma were 60% and 80%, and the median PFS and OS were 6.6 and 13.8 months, respectively. Subgroup analysis highlighted a 69% ORR in PD-L1-positive patients and 38% in PD-L1-negative patients [
99] (Table
2). This small sample study suggests that chemotherapy combined with anti-PD-1 has potential in gastric or gastroesophageal conjunctive adenocarcinoma, although confirmatory findings are required. In a related follow up, an investigation of the efficacy of chemotherapy combined with PD-1 blockades, KEYNOTE-062, is in progress to assess this combination as a first-line therapy for advanced gastric or gastroesophageal junction adenocarcinoma.
The preliminary results of a phase I/II study of ramucirumab plus nivolumab in patients with previously treated advanced gastric adenocarcinoma found a partial response was obtained in ten patients, representing a 22% of the study population with a DCR of 59% [
100]. In addition, a phase I study (NCT02443324), which assessed the efficacy of pembrolizumab in combination with ramucirumab, found a 50% DCR and PD-L1-positive patients appear to have significantly benefited [
111]. Combination immunotherapies in esophageal and gastric cancer have achieved a preliminary advantage, and sequencing combination therapies is also moving forward.
Hepatobiliary carcinoma
Presently, targeted drugs, such as sorafenib, lenvatinib, and regorafenib, are the primary treatments for advanced hepatocellular carcinomas (HCC). Recent results have indicated the potential of PD-1/PD-L1 blockades for the treatment of advanced HCC. In the CheckMate-040 study, the overall ORR of the patients administered with nivolumab was 14–23%. Subgroup analysis suggested that the DCR in patients without sorafenib was 54% with an OS of 28.6 months. In patients treated with sorafenib, the ORR was 55%, suggesting that there may be only a fractional benefit, although this group had a prolonged 15.6 month OS [
113]. In addition, liver toxicity of PD-1/PD-L1 blockades was lower than that of conventional drugs. As a result in 2017, nivolumab was approved by the FDA as a second-line treatment for HCC. Preliminary results from the KEYNOTE-224 study are similar to those of CheckMate-040, the ORR, and DCR in patients with advanced HCC whom had previously been treated with sorafenib was 17% and 61%, respectively [
114]. In view of the aforementioned findings, the phase III CheckMate-459 trial which will compare nivolumab with sorafenib as first-line treatments for advanced HCC with overall survival as the primary endpoint is much needed [
115].
PD-1 inhibitor monotherapies appear to be well tolerated with relatively consistent efficacy in liver cancer patients. For example, the retrospective study of CheckMate-040 trial found a 50% ORR in 14 patients whom had received nivolumab combined with local-regional treatment with three CRs (11%) and five PRs (18%) [
116]. To further increase the antitumor response, a preliminary study of lenvatinib plus pembrolizumab in patients with unresectable HCC resulted in encouraging antitumor activity and tolerance with 46% ORR (Table
3). The most common AEs were decreased appetite and hypertension without new safety signals [
117].
Table 3
Key trials of combination immunotherapy in hepatocellular, biliary tract, and pancreatic cancers
CheckMate-040 retrospectively evaluate; | Advanced HCC | | 28 | Nivolumab + local-regional treatment | PD-1 + LR | ORR: 50%; SD: 21%; mOS: 13.6 months | Grade 3: 7% | |
NCT03006926 | Unresectable HCC; | Ph-1b | 13 | Lenvatinib + pembrolizumab | PD-1 + Target | ORR: 46%; SD: 46%; | Any grade: 94%; decreased appetite: 56%; hypertension: 56% | |
NCT02715531 | Unresectable or metastatic HCC | Ph-1b | 68 | Atezolizumab + bevacizumab | PD-L1 + Target | ORR: 34%; PFS rate at 6 months: 71% | Any grade: 72%; Grade 3/4: 25% | |
Study-022; NCT02519348 | Advanced HCC | Ph-1/2 | 40 | Durvalumab + tremelimumab | PD-L1 + CTLA-4 | ORR: 18%; DCR: 57.5% | Any grade: 72%; Grade 1–3: 20% | |
JapicCTI-153098 | Biliary tract cancer | Ph-1 | 30 | Nivolumab 240 mg, 2-week intervals + cisplatin-gemcitabine | PD-1 + Chemo | ORR: 36.7%; mPFS: 4.2 months; mOS: 15.4 months | Malaise (8/30, 27%) and decreased appetite (7/30, 23%) | |
2018 ASCO Poster | Advanced intrahepatic cholangiocarcinoma | 14 | Lenvatinib + pembrolizumab or nivolumab | PD-1 + Targeted | ORR: 21.4%; DCR: 92.9%; mPFS:5.9 months | Grade 3: 14% | |
NCT01938612 | Biliary tract cancer | Ph-1 | 65 | Durvalumab 20 mg/kg + tremelimumab 1.0 mg/kg, q4w; durvalumab monotherapy | PD-L1 + CTLA-4 | DCR: D, 16.7%; D + T, 32.2%; mPFS: D, 9.7 months, D + T, 8.5 months; mOS: D, 8.1 months; D + T, 10.1 months | Any grade: D, 64%; D + T, 82%; Grade ≥ 3: D, 19%; D + T, 23%; D + T: a death due to drug-induced liver injury | |
NCT02821754 | Advanced HCC; advanced BTC | Ph-2 | 22 | Monthly tremelimumab 75 mg + durvalumab 1500 mg for 4 doses followed by monthly durvalumab 1500 mg monotherapy | PD-L1 + CTLA-4 | ORR: HCC, 20%; BTC, 0%; DCR: HCC, 60%; BTC, 41.7%; mPFS: HCC, 7.8 months, nivo alone, 3.1 months; mOS: HCC, 15.9 months; BTC, 5.45 months | Grade 3/4: lymphocytopenia, hyponatremia, bullous dermatitis, maculopapular rash | |
KEYNOTE-202; NCT02826486 | Metastatic pancreatic adenocarcinoma | Ph-2a | 37 | BL-8040 + pembrolizumab | PD-1 + Molecules | PR: 3.4%; DCR: 34.5%; mOS: 3.4 months; OS rate at 6 months: 34.9% | |
NCT02309177 | Advanced pancreatic cancer | Ph-1 | 50 | Nab-paclitaxel 125 mg/m2 + gemcitabine 1000 mg/m2 on day 1, 8, and 15 + nivolumab 3 mg/kg on d 1 and 15 of each 28-day cycle | PD-1 + Chemo | ORR: 18%; DCR:64%; mPFS: 5.5 months; mOS: 9.9 months | Grade 3/4: 96%; Grade 5: 1 pts | |
NCT02311361 | Advanced pancreatic adenocarcinoma | Ph-1/2 | 51 | Cohort 1: Durvalumab + SBRT; Cohort 2: SBRT followed by durvalumab; Cohort 3: Durvalumab + Tremelimumab + SBRT; Cohort 4: SBRT followed by Durvalumab + Tremelimumab | PD-L1 + CTLA-4 + RT; PD-L1 + RT | PR: cohort 1, 1 pt.; cohort 4, 2 pts.; mPFS and mOS: cohort 1, 1.7 months, 3.4 months; cohort 2, 2.6 months, 9.1 months; cohort 3, 1.6 months and 3.0 months; cohort 4, 3.2 months, 6.4 months | The most commonly TRAEs were lymphopenia. Grade 3–4: lymphopenia and anemia | |
The FDA recommends atezolizumab combined with bevacizumab as a first-line therapeutic regimen for patients with advanced HCC based on a phase 1b study (NCT02715531). The findings of this study highlight a 34% ORR associated with atezolizumab combined with bevacizumab among 68 patients assessed [
118], although this was a relatively small study. The recent phase III IMbrave150 trial is based upon these findings and will evaluate the efficacy and safety of this combination compared to sorafenib in participants with locally advanced or metastatic HCC who have received no prior systemic treatment [
127]. Combining PD-1 blockade and CTLA-4 blockade for advanced HCC may also prove beneficial and early data from NCT02519348 suggests relative safety with an 18% ORR [
113] and the upgraded study is currently recruiting. In addition, several clinical trials of PD-1/PD-L1 blockades combined with other types of antitumor therapy are also under way.
Related basic medical research by Nakamura et al. divided biliary tract cancers (BTC) into four molecular subgroups based upon prognostic gene profiles and found that classification correlates with patient prognosis. Among subtypes with the worst prognosis, the expression of immune checkpoint-related molecules, including PD-L1, was upregulated more than in any other subgroups, which again suggests immune checkpoint inhibitors may yield a favorable response [
128]. In addition, emerging data suggests MMR or MSI-H mutation tumors have a much higher response rate to PD-1/L1 inhibitors, and in cholangiocarcinoma, MSI-H accounting for 5% of gallbladder cancers (GBC), 5–13% of extrahepatic cholangiocarcinoma (ECC), and 10% of intrahepatic cholangiocarcinoma (ICC) [
109]. Phage 1b KEYNOTE-028 trail assessed the safety and activity of pembrolizumab monotherapy among advanced solid tumors with PD-L1 expression ≥ 1%, and the cholangiocarcinoma cohort suggested that of 24 patients who met the evaluation criteria ORR was 17% [
129].
Sequencing exons and transcriptomes has revealed heterogeneous molecular changes among cholangiocarcinoma, and the selection of an immunotherapy combined with a targeted therapy may provide answers where other avenues may not. One small sample study found after treatment with PD-1 blockades combined with lenvatinib, 3:14 patients had a 21.4% ORR and a 93% DCR. Interestingly, this study using 450-gene next generation sequencing (NGS) panel in seven patients to detect all classes of genetic status discovered that having a high TMB might be used to indicate preferential treatment [
121] (Table
3). The standard first-line chemotherapy for advanced BTC is gemcitabine plus cisplatin; however, there is no standardized second-line intervention. This is because evidence is lacking to guide specialists. PD-1/L1 blockades combined with a standard chemotherapy is frequently administered as a second-line therapy, although there appears to be an element of trial and error adjustment. Currently, several clinical trials are under way, including one investigating a guadecitabine and durvalumab combination (NCT03257761) and another pembrolizumab and FOLFOX (NCT02268825) (Table
3). The findings of these studies may provide support for clinicians seeking the most effective option where first-line treatments have failed.
Another interesting research avenue which has emerged is around the impact of standards of care (SoC). Currently under way, a phase III clinical study is exploring this in more detail, focusing on the efficacy of PD-1 blockade combined with SoC compared with SoC alone for the treatment of previously untreated locally advanced or metastatic BTC. The primary hypothesis of the study is that participants will have a longer OS when treated with combined therapy than when treated with SoC alone, although this study may also provide insight into the interactions between SoC and PD-1 blockades which is also needed.
Pancreatic carcinoma
Previously presented evidence suggests that immunotherapy combined with PD-1/PD-L1 blockades may deliver favorable outcomes with durable responses for various types of cancer; however, pancreatic carcinomas remain refractory. Except for MSI-positive pancreatic cancers which accounts for approximately 1.2%, the efficacy of PD-1/PD-L1 blockades alone are unsatisfactory for most pancreatic cancers. Unfortunately, more than 10% of patients develop grade 3 and 4 AEs, which is likely to be at least partly be due to the unique microenvironments (TME) in the pancreas [
130]. Pancreatic tumor tissues are characterized by excessive cancer-associated fibroblasts (CAFs), dense connective tissue, low vascular density, and insensitivity to ischemia and hypoxia. In addition, immunosuppressive immune cells, such as M2 macrophages, are found in tumor tissues which inhibit immune killer cells from effectively entering through the tumor matrix [
131]. Potentially, combined immunotherapies could provide a solution to these problems by bolstering the immune response to pancreatic tumor development.
Presently, gemcitabine, albumin paclitaxel, and a monoclonal CD40 antibody combined with nivolumab are frequently used as pancreatic cancer interventions. These interventions act by destroying tumor matrices and by exposing more antigens, which promote lymphocyte infiltration. Cabiralizumab (FPA008) is an anti-CSF-1R antibody which can cause the depletion of tumor-associated macrophages (TAMs) which may provide additional benefit. As such, one recent study (NCT02526017) was designed specifically to evaluate the safety, tolerability, as well as the clinical benefit of cabiralizumab in combination with nivolumab in patients with selected advanced cancers, including pancreatic cancer. The study revealed lasting clinical benefit for five patients with advanced pancreatic cancer who were insensitive to a previously administered single-drug immunotherapy, including three patients with microsatellite stability (MSS). However, the sample size of the study was small (n = 33), therefore these results ought be verified under stricter conditions, including a larger sample size based on a pre-trial calculation using best available evidence, and with an appropriate control group. Importantly, it is necessary to conduct this research focusing on those suffering pancreatic cancer specifically because of the refractory nature of this condition but also to explore therapeutic effects across stages.
A phase II clinical trial (NCT03336216) currently under way is focusing on the efficacy of cabiralizumab and nivolumab combined with or without chemotherapy specifically for the treatment of advanced pancreatic cancer. Chemotherapy in this particular trial includes paclitaxel, gemcitabine, irinotecan, or FOLFIRINOX. The researchers have proposed to recruit 160 patients which is substantially larger than previously mentioned NCT02526017 study, and to use PFS as the primary clinical endpoint. The potential benefit of PD-1/PD-L1 blockades combined with other therapeutic approaches has resulted in a number of trials focusing on resectable pancreatic cancer, broad line resectable pancreatic cancer, and advanced pancreatic cancer. Most of the trials being designed are again preclinical studies or early phase clinical research but hopefully findings from the aforementioned studies will develop this evidence base and drive higher level clinical research.
Colorectal carcinoma
The KEYNOTE-028 trial which involved a cohort of people with existing colon and rectum carcinomas found only a 4% ORR for pembrolizumab monotherapy after screening out patients with PD-L1 > 1% (
n = 1), and there was no significant improvement when compared with that of unscreened patients [
132]. DMMR/MSI-H-type mCRC accounts for 4% of mCRC overall, although this is insensitive to traditional chemotherapy and generally has a poor prognosis. However, many neoantigens increase dMMR patients’ sensitivity to PD-1/PD-L1 blockade therapy. Therefore, nivolumab has been approved for patients with metastatic DNA mismatch repair-deficient colorectal cancer based on the Checkmate 142 study suggesting 23 of 74 patients achieved objective response and 68.9% of patients had disease control for ≥ 12 weeks [
133].
Nevertheless, the colorectal cancer group of phase II clinical trials evaluating the clinical activity of pembrolizumab in patients with progressive metastatic carcinoma has shown that the ORR and DCR of patients with mismatch repair-deficient (dMMR) within 20 weeks were 40 and 90%, respectively. For the mismatch repair-proficient (pMMR) group, these values were 0 and 11%, respectively which suggests that mismatch repair status may be used as efficient indicators of PD-1 antibodies, although further research is needed for clarification [
109]. One phase 3 clinical trial (NCT02563002) has been designed to investigate these issues and will compare PFS and OS between dMMR/MSI-H patients administered single-drug PD-1 inhibitor therapy and dMMR/MSI-H patients administered standard chemotherapy.
Concerning double immunotherapy in dMMR/MSI-H mCRC, results for the nivolumab plus ipilimumab cohort of CheckMate-142 study found at the median follow-up (13.4 months) a 55% ORR with corresponding PFS and OS rates at 12 months of 76% and 87%, respectively [
134] (Table
4). Therefore, indirect comparisons suggest that combination therapies provide improved efficacy relative to anti-PD-1 monotherapy (ORR 31%) and has a favorable benefit-risk profile. Importantly, the study also suggests that there is no relationship between efficacy and the expression of PD-L1 in MSI-H patients.
Table 4
Key trials of combination immunotherapy in colorectal cancers
CheckMate-142; NCT02060188 | DNA mismatch repair-deficient/microsatellite instability-high metastatic colorectal cancer | Ph-2 | 119 | Nivolumab 3 mg/kg + ipilimumab 1 mg/kg q3w, followed by nivolumab 3 mg/kg once q2w | PD-1 + CTLA-4 | ORR: 55%; DCR: 80%; PFS rate at 12 months: 71%; OS rate at 12 months: 85% | Grade 3/4: 32% | |
NCT01633970 | Metastatic colorectal cancer | Ph-1 | 44 | Arm A, MPDL3280A (anti-PDL1) + bevacizumab; Arm B, MPDL3280A + bevacizumab + FOLFOX | PD-L1 + Target; PD-1 + Target + Chemo | ORR: Arm A, 8%; Arm B, 36%; | Grade 3/4: Arm A, 64%; Arm B, 73% | |
NCT01988896 | Metastatic colorectal cancer | Ph-1b | 84 | Atezolizumab + cobimetinib | PD-L1 + Target | ORR: 8%; DCR: 31%; OS rate at 12 months: 43%; mOS: 9.8 months; PFS rate at 6 months: 18% | Any grade: 96%; Grade 3/4: 32% | |
NCT02375672 | Colorectal cancer irrespective of MMR status | Ph-2 | 40 | Pembrolizumab + mFOLFOX6 | PD-1 + Chemo | ORR: 53%; DCR at 8 weeks: 100% | Grade 3/4: 36.7% | |
NCT02437071 | Mismatch repair proficient (pMMR) metastatic colorectal cancer | Ph-2 | 26 | Pembrolizumab + radiotherapy; pembrolizumab + ablation | PD-1 + RT; PD-1 + LR | ORR: RT cohort, 9%; LR cohort, no responses | Grade 1/2: 73% | |
NCT02981524 | Mismatch repair–proficient (MMR-p) advanced colorectal cancer | Ph-2 | 17 | Pembrolizumab + CyGVAX colon vaccine + cyclophosphamide | PD-L1 + vaccine + chemo | ORR: 18%; mPFS: 2.7 months; mOS: 7.0 months | Two patients (12%) had grade 3/4 adverse events that were attributed to study therapy | |
NCT02870920 | Advanced refractory colorectal carcinoma, not selected for dMMR | Ph-2 | 180 | Durvalumab 1500 mg D1 q 28 days + tremelimumab 75 mg D1 for first 4 cycles vs BSC | PD-1 + CTLA-4 | DCR: D + T, 22.7%; BSC, 6.6%; mOS: D + T, 6.6 months, BSC, 4.1 months | Grade 3/4: abdominal pain, fatigue, lymphocytosis and eosinophilia were significantly higher in D + T | |
As mentioned previously, PD-1 inhibitor monotherapy has little effect in patients with microsatellite stable colorectal cancer. Indeed, many factors may influence the efficacy of PD-1/PD-L1 blockade in patients with colorectal cancer, including gene mutations, the immune microenvironment, and a patient’s genetic inheritance. In unscreened patients with advanced colorectal cancer, a small sample study at the 24-week follow-up found 53% ORR for PD-1 blockade combined with chemotherapy. Although, it remains unclear how effective chemotherapy alone will be for this group of patients due to the lack of rigorous experimental design, and the proportion of people (36.7%) suffering associated severe side effects associated [
135].
MEK inhibition upregulates tumor major histocompatibility complex-I expression, promoting intra-tumoral T cell accumulation while improving anti-PD-L1 responses [
140]. For patients with MSS colorectal cancer, recent studies have found that cobimetinib (MEK1/2 inhibitor) combined with PD-L1 blockades results in a DCR of 31%, and 43% of patients survive for more than 12 months [
136]. As a result, a phase III clinical trial (NCT02788279) was designed to evaluate atezolizumab in combination with cobimetinib versus atezolizumab or regorafenib monotherapies and the findings are eagerly anticipated.
An increasing number of clinical trials are currently under development and ongoing which provides some optimism. However, these combinations face a number of problems, such as the need for more comprehensive gene sequencing and the difficulty of accurately and rigorously classifying colorectal cancer patients to predict treatment efficacy. In addition, the use of the same treatment regimen for different patients may not improve prognoses due to significant differences among individual patients which suggest the need for personalized cancer care. However, in order for this to become a reality, studies need to be scaled up and studies ought to be designed to incorporate the subtle differences between participants, which one could argue is not the current state of play.
Conclusions and perspectives
The advantages of combined immunotherapy based on PD-1/PD-L1 blockades for various tumors appear to be the logical next step. Although, there are a great number of unknowns, including dose/response, safety, tolerability, durability, and indeed efficacy. How these new treatment options will be placed within the existing treatment framework is a concern. Researchers are endeavoring to answer these questions through rigorous clinical trials focusing on specific types of tumors and within specific populations at various stages of these diseases. Studies have found an increase in the proportion of immune-related adverse events after receiving combination therapy compared to monotherapies. Although, these generally include diarrhea, fatigue, and hypothyroidism, which are within a tolerable range and manageable [
17].
The increasing the number of combination studies has highlighted beneficial antitumor effects in early clinical stages. However, results from several clinical trials found no enhanced benefit for the patients with advanced cancers. Moreover, administering combination immunotherapies has been found to increase treatment toxicity. In patients who received radiotherapy prior to treatment with PD-1 blockades, research has revealed that immune inflammation frequently and naturally recurs at the original site of irradiation. Therefore, as many of the current combined immunotherapeutic methods remain experimental, developing this evidence base is absolutely necessary.
Understanding the underlying mechanisms of each therapeutic combination as well as the subtleties of individual responses is required to avoid combination schemes which do harm. Ironically, combination immunotherapeutic models pose similar questions to traditional treatment: What is the ideal patient population for which combination? Is the required combination therapy sequential or concurrent? What timing and adjustment criteria can be used for continuous and combined interventions? What is the related safety and toxicity of each combination? All of these questions require a sophisticated evidence-base developed through mature theoretical foundations and basic medical research. Once small sample studies have been conducted, larger studies ought to be commenced as is currently occurring. However, at present, it would appear as though we are trying to improve outcomes by combining a possible best available treatment with a potential catalyst or less subtly, simply seeking compatible combinations. We must not overlook the fact that this is essentially combining an average of averages with yet another. More specific research is required with more comprehensive data collection if we are to treat individuals with more precision and sensitivity as is required for gastrointestinal malignancies. Further research should focus on markers as these may provide measurable trajectories to accurately predict the benefit of combination therapies.