Japan Society of Clinical Oncology provisional clinical opinion for the diagnosis and use of immunotherapy in patients with deficient DNA mismatch repair tumors, cooperated by Japanese Society of Medical Oncology, First Edition

In Japan, approximately 380,000 people die of malignant neoplasm (cancer) annually, and cancer is the number one cause of death. Improving the outcome of cancer treatment is a critical issue for the Japanese public. In the field of cancer pharmacotherapy, the advent of effective novel therapeutic drugs has improved treatment outcomes and prognoses. In parallel, the development of biomarkers to identify patients for whom a certain treatment is expected to be effective before starting treatment has contributed to the improvement of cancer treatment outcomes.

In December 2018, in Japan, pembrolizumab, a PD-1 inhibitor, was approved for advanced/recurrent MSI-H solid tumors. This is the first drug in Japan for tumor-agnostic indications. This treatment is expected to be a novel treatment option for solid tumors that are difficult to cure, while there are some issues related to administering the treatment in the clinical setting:

For the issues described above, the various clinical practice guidelines published to date only briefly describe key points in the use of immune checkpoint inhibitors in patients with dMMR solid tumors. Since no comprehensive guidelines cover all key points regardless of primary tumor site, it is important to integrate common, tumor-agnostic views to the extent possible and provide a guide for clinical care to prevent confusion at clinical sites.

The current guidelines systematically describe items to be considered when seeing patients with dMMR solid tumors, including the timing and methods of testing defective mismatch repair function, the positioning of PD-1/PD-L1 inhibitor therapy, and clinical care systems. Moreover, given that recent progress in analytical techniques is facilitating rapid development of comprehensive genetic testing methods using next-generation sequencing and somatic cell genetic testing methods using blood samples (liquid biopsy), these novel testing methods are also included. In the clinical setting in Japan, if appropriate tests are performed on appropriate patients and the patients receive appropriate treatment at appropriate timing based on the recommended levels described in the present guidelines, treatment outcomes in patients with solid tumors are expected to be improved.

In the preparation of the guidelines, clinical questions (CQ) were formulated, and evidence for answers to the CQs was gathered by handsearch. Based on the search results, the committee members voted to determine a recommended level for each CQ (Table 1). The recommended levels were determined by taking into account the strength of evidence for each CQ, expected benefits and losses of patients, and other factors. In voting, whether the contents of medical care (including tests and indications) are approved or covered by health insurance in Japan was not considered. However, relevant information was described in the remarks column as needed. The committee’s opinions were determined in the following manner: (1) if SR accounted for at least 70% of the vote, the committee’s opinion was SR; (2) if (1) was not met, but SR + R accounted for at least 70% of the vote, the committee’s opinion was R; (3) if (1) or (2) was not met, but SR + R + ECO accounted for at least 70% of the vote, the committee’s opinion was ECO; (4) if NR accounted for at least 50% of the vote, the committee’s opinion was NR, irrespective of the results of (1)–(3); and (5) if none of (1)–(4) was met, there was “no recommended level.”

At present, some recommendations for CQs are not based on sufficient evidence. It is also possible that the accumulation of new evidence in the future will lead to substantial changes in the descriptions in the text and recommended levels. Consequently, the guidelines are positioned as a “provisional clinical opinion,” taking into account that the guidelines contain many recommendations made based on a consensus among the committee members at the current level.

Based on the results of a pooled analysis of 149 patients with advanced/recurrent dMMR solid tumors that progressed after chemotherapy from five clinical studies of pembrolizumab [KEYNOTE-016 study, KEYNOTE-164 study (cohort A), KEYNOTE-012 study, KEYNOTE-028 study, and KEYNOTE-158 study), the United States Food and Drug Administration (FDA) approved pembrolizumab for dMMR solid tumors including colorectal cancers that are resistant to standard systemic treatment or for which no standard treatment is available, on May 23, 2017. In Japan, pembrolizumab was approved on December 21, 2018, based on the updated results of the KEYNOTE-164 study (cohort A) and KEYNOTE-158 study (Table 12).

A study of nivolumab monotherapy and nivolumab/ipilimumab (an anti-CTLA4 antibody drug) combination therapy in patients with dMMR colorectal cancers (the CheckMate-142 study) reported good outcomes with the ORRs of 31% and 55%, respectively, and the median PFSs was not reached in either group [60, 61]. A therapeutic effect was observed irrespective of the degree of PD-L1 expression, the presence of the BRAF/KRAS mutations, and the presence of Lynch syndrome. Patient evaluation using EORTC QLQ-C30 demonstrated improved QOL and clinical symptoms [60, 61]. Based on these results, the FDA approved nivolumab monotherapy in August 2017 and nivolumab/ipilimumab combination therapy in July 2018 for metastatic dMMR colorectal cancers that progressed after treatment with fluoropyrimidine, oxaliplatin, and irinotecan. For durvalumab, a PD-L1 inhibitor, a phase II study in patients with dMMR colorectal cancers and phase I/II studies in patients with dMMR solid tumors were conducted and demonstrated an efficacy with the ORR for colorectal cancers of 22% and an overall ORR of 23% [62]. Efficacy for dMMR solid tumors was reproduced in case reports and the analyses of dMMR subgroups in prospective phase II studies.

As for dMMR colorectal cancer, the KEYNOTE-164 study reported good outcomes not only in patients who had received chemotherapy with fluoropyrimidines, oxaliplatin, and irinotecan hydrochloride hydrate (cohort A), but also in 63 patients who had received one or more regimens of chemotherapy (cohort B) with the ORR of 32% (95% CI 21–45), the median PFS of 4.1 months (95% CI ≥ 2.1, upper limit not reached), and the median OS not reached. Therefore, the use of pembrolizumab in second- or later-line treatment is considered. Moreover, a phase III study comparing standard systemic treatment and pembrolizumab therapy in patients receiving first-line treatment is underway. If this study demonstrates the efficacy of pembrolizumab in first-line treatment for dMMR colorectal cancers, dMMR testing before the start of first-line treatment would be desirable.

The efficacy of PD-1/PD-L1 inhibitors has been confirmed consistently in dMMR solid tumors, although these reports did not have a sufficient number of patients by cancer type or by treatment line. Molecular biology also suggests a commonly high immunogenicity in dMMR solid tumors. As for adverse events, although caution is needed for the serious immune-related adverse events that often occur, they are generally tolerable. Therefore, for all patients with dMMR solid tumors, including tumors for which PD-1/PD-L1 inhibitors have no approved organ-specific indications from the viewpoint of efficacy and safety, PD-1/PD-L1 inhibitors can be a potent treatment option. The previous clinical studies were conducted in patients who had difficulty receiving standard systemic treatment (including patients with treatment resistance, intolerance due to adverse events, and not treated at patients’ request). When cancer progresses, the patient’s general condition is often worsened. Considering the TAT of dMMR testing, it is desirable to perform dMMR testing early to determine eligibility for PD-1/PD-L1 inhibitors.

Based on the above considerations, for patients with solid tumors who are receiving standard systemic treatment or who have difficulty receiving any standard treatment, dMMR testing is highly recommended to determine eligibility for PD-1/PD-L1 inhibitors.

As of April 2019, Table 13 shows the types of solid tumors for which PD-1/PD-L1 inhibitors can be used in clinical practice or are expected to be used in the future (as of April 2019).

For solid tumors for which PD-1/PD-L1 inhibitors can be used in second- or later-line treatment irrespective of MMR function, the applicability of PD-1/PD-L1 inhibitors is judged irrespective of MMR function. Therefore, in principle, it is not necessary to perform dMMR testing. For gastric cancer, nivolumab therapy is recommended in third- or later-line treatment irrespective of the presence of microsatellite instability, but only for dMMR cancer, the guidelines recommend the use of the therapy in second- or later-line treatment [63]. Thus, if the treatment line of PD-1/PD-L1 inhibitors is expected to become earlier depending on MMR function, administration of dMMR testing is also considered.

If there is a solid tumor for which the applicability of PD-1/PD-L1 inhibitors is judged based on a biomarker other than the dMMR status such as PD-L1 expression and the biomarker is negative, dMMR testing is recommended, because PD-1/PD-L1 inhibitors are expected to be effective if the tumor is dMMR, as shown in Fig. 6.

For malignant melanoma, PD-1 inhibitors have demonstrated efficacy as adjuvant therapy and have been approved (KEYNOTE-054 study [64] and ONO-4538-21 study [65]). For non-small cell lung cancer, durvalumab, a PD-L1 inhibitor, has been approved based on the results of the PACIFIC study, a randomized, double-blind, placebo-controlled, multicenter phase III study of durvalumab administered sequentially in patients with unresectable locally advanced cancer (stage III) who did not show disease progression after curative concurrent chemoradiotherapy (CRT) using platinum drugs [66]. However, since no difference in efficacy due to MMR function has been reported from these studies, dMMR testing before treatment is not necessary in principle. For other solid tumors, the efficacy of immune checkpoint inhibitors as perioperative treatment has not been established. Therefore, if the tumor is curable with local therapy, dMMR testing to select therapeutic drugs is not necessary in principle. Thus, at present, for patients with solid tumors that are not locally advanced or metastatic, dMMR testing for determining eligibility for PD-1/PD-L1 inhibitors is not recommended.

However, it is known that dMMR is a favorable prognostic factor for colorectal cancer, particularly for stage II colon cancer, and if the cancer is dMMR, adjuvant therapy with fluoropyrimidines is unnecessary. Therefore, it is considered to be desirable to perform dMMR testing to judge the necessity of adjuvant chemotherapy (for details, refer to “Guidance on Genetic Testing in the Clinical Practice of Colorectal Cancer, Third Edition”). Moreover, currently, a study to verify the efficacy of perioperative use of immune checkpoint inhibitors and a study to concurrently use immune checkpoint inhibitors and chemoradiotherapy for locally advanced cancer are underway. If good outcomes are obtained from these studies, dMMR testing will be necessary for solid tumors curable with local therapy.

For some solid tumors, PD-1/PD-L1 inhibitors have been approved irrespective of MMR function. The effectiveness of a PD-1/PD-L1 inhibitor in patients who have already received another PD-1/PD-L1 inhibitor has not been demonstrated. Therefore, dMMR testing for the purpose of administration of PD-1/PD-L1 inhibitors in patients with solid tumors who have already received a PD-1/PD-L1 inhibitor is not recommended.

Although the frequency of dMMR is high (80–90%) in Lynch-associated colorectal cancers [67], not all tumors that develop in patients with Lynch syndrome have dMMR. Because the efficacy of PD-1/PD-L1 inhibitors is influenced by the MMR function of the tumor, PD-1/PD-L1 inhibitors are not expected to be effective for pMMR tumors even in patients with Lynch syndrome. Therefore, dMMR testing for determining eligibility for PD-1/PD-L1 inhibitors is also recommended for tumors that develop in patients with Lynch syndrome.

The pooled analysis of patients with dMMR from five KEYNOTE studies (KEYNOTE-016 study, KEYNOTE-164 study (cohort A), KEYNOTE-012 study, KEYNOTE-028 study, and KEYNOTE-158 study) that enrolled patients who were determined to be dMMR based on IHC or MSI testing performed at each study site demonstrated good anti-tumor effect of pembrolizumab. Among 149 patients, 60 patients were determined to be dMMR by MSI testing alone, 47 patients by IHC alone, and 42 patients by both tests [68]. Among them, only 14 patients were determined to be MSI-H by MSI testing performed at a central testing laboratory. A phase II study of nivolumab in patients with colorectal cancer who were determined to be dMMR (the CheckMate-142 study) enrolled patients who were determined to be dMMR by IHC or MSI testing performed at each study site and has demonstrated the efficacy of nivolumab [60]. Thus, if a cancer is determined to be dMMR by either IHC or MSI testing, it is eligible for PD-1/PD-L1 inhibitors, although there may be some differences depending on the type of cancer.

In Japan, in September 2018, “MSI test kit (FALCO)” was approved as a companion diagnostic for pembrolizumab. Any institution in Japan can order this test, and the test is performed in quality-assured testing facilities. Moreover, this test kit can determine the dMMR status by testing tumor tissue alone if tumor cells account for ≥ 40% of the tumor tissue, which is, therefore, very convenient [45]. Thus, as a dMMR-testing method for determining eligibility for PD-1/PD-L1 inhibitors, MSI testing is highly recommended.

As mentioned above, the efficacy of immune checkpoint inhibitors was demonstrated in patients enrolled in the pooled analysis of five KEYNOTE studies and those in the Checkmate-142 study, who were diagnosed as having dMMR based on IHC or MSI testing performed at each study site. In both analyses, the efficacy of PD-1 inhibitors was demonstrated also in patients who were determined to be dMMR by IHC alone. Actually, in the Checkmate-142 study, in which MSI was determined centrally by MSI testing (with 5 markers used in the Bethesda panel and TGFR type 2), 14 of the 74 patients who were determined to be dMMR at each study site were judged to be non-MSI-H. However, 3 of the 14 patients (21%) responded to treatment [60], and this fact suggests that even when the results of the two tests are not consistent and the dMMR was diagnosed based only on one test, the anti-tumor effect of immune checkpoint inhibitors can be expected. Compared to MSI testing and NGS testing, IHC can be performed inexpensively at individual medical institutions. However, there are some issues. More specifically, as of March 2019, no antibody for IHC has been approved as an in vitro diagnostic in Japan; there are variations in staining depending on the antibodies and staining conditions, and the evaluation method has not been well established. Consequently, IHC is recommended as a dMMR-testing method for determining eligibility for PD-1/PD-L1 inhibitors (however, as of March 2019, no antibody for IHC has been approved as an in vitro diagnostic in Japan).

While a high concordance rate between MSI testing results and IHC results has been reported, some inconsistent cases have been reported. One example is pathogenic missense variants of the MMR genes [69, 70]. In this case, proteins that have lost MMR function are expressed. Therefore, the MSI testing result indicates MSI-H and the tumor is determined to be dMMR, while in IHC, MMR proteins are detected, and the tumor is determined to be MMR proficient (false negative). For this dMMR tumor, PD-1/PD-L1 inhibitors are presumed to be effective. It has been reported that such missense variants are observed in approximately 5% of patients with Lynch syndrome [71]. On the other hand, possible causes of false-negative cases by MSI testing include a low tumor cell ratio. Actually, a tumor cell ratio of ≥ 50% is recommended for the MSI test (FALCO). The positive predictive value of IHC or MSI testing has been reported to be 90.3% [72]. It has been reported that when patients who were diagnosed with dMMR solid tumors by IHC or MSI testing and received PD-1/PD-L1 inhibitors, but did not respond to the therapy were evaluated again by both MSI testing and IHC, 60% of them were found to be MSI-L/MSS/pMMR [72]. To extensively identify patients who can benefit from PD-1/PD-L1 inhibitors, testing should be performed based on a good understanding of the characteristics of both tests. If a false-positive or false-negative result is expected or if there are doubts about the precision or results of the test, performing the other test should be considered.

In Japan, on December 27, 2018, the FoundationOne CDx received marketing approval for obtaining comprehensive cancer genome profiles of a tumor tissue from patients with solid tumors and for detecting somatic cell genetic alterations to determine the applicability of some molecular targeted drugs.

Because FoundationOne CDx includes MSI testing using the NGS method, the comprehensive cancer genome profiling and MSI testing (the NGS method) can be performed simultaneously for each cancer type with specimens and at the timing specified in the latest guidelines and other documents issued by relevant academic societies. However, as of March 2019, dMMR testing using the FoundationOne CDx is not covered by health insurance, and there are requirements for facilities to perform the FoundationOne CDx. Therefore, dMMR determination using the NGS method can be accessed at limited facilities in Japan. The FoundationOne CDx also has problems in feasibility. More specifically, it has a certain level of failure rate and needs a large amount of DNA for analysis.

In the five KEYNOTE studies and the Checkmate-142 study conducted for the application for the FDA approval of pembrolizumab, screening tests for dMMR did not include NGS testing. However, the determination of MMR function using NGS testing and MSI testing has a similar measurement principle in that a repeat number of microsatellites is used to determine whether a tumor is dMMR, and it has been reported that the concordance rates between these tests were extremely high, 99.4% in colorectal cancers, and 96.5% in solid tumors other than colorectal cancers [73]. Moreover, when inconsistent cases were analyzed, they were dMMR by IHC, suggesting that NGS testing is more useful. Therefore, it is scientifically unnecessary to perform testing using the MSI test kit (FALCO), a companion diagnostic, or IHC to reconfirm the status determined to be MSI-H by NGS testing, for which analytical validity has been established in the determination of MSI. Thus, an NGS testing approach for which analytical validity has been established is recommended as a dMMR-testing method for determining eligibility for PD-1/PD-L1 inhibitors.

[Note: Liquid biopsy test]

The usefulness of liquid biopsy, which uses body fluid samples such as blood and urine to diagnose the condition of a tumor instead of directly using tumor tissues, has also been reported. The blood usually has a certain amount of free DNA, but the amount of free DNA increases in cancer patients. DNA present in plasma, regardless of whether it is from normal cells or tumor cells, is called cell-free DNA (cfDNA). Because cfDNA in a cancer patient contains DNA from tumors, it is often called circulating tumor DNA (ctDNA). Studies that verified tumor tissues and ctDNA using the MSI test kit and NGS testing reported high sensitivity (86–100%) and specificity (99–100%) [74, 75]. If no tumor tissue is available for testing, therefore, a test using ctDNA is expected to detect genetic alterations in tumor cells in a minimally invasive manner and in real time.

[Note: Relationship between TMB/PD-L1 and MMR]

As biomarkers for the efficacy of PD-1/PD-L1 inhibitors, MSI-H, tumor mutation burden-high (TMB-H), and PD-1/PD-L1 protein expression have been reported.

The proportion of these factors (biomarkers) varies among different cancer types and one factor can confound other factors. In a report of the study that verified the associations among MSI (by NGS), TMB, and PD-L1 protein expression in 11,348 patients with solid tumors, the frequency of the factors and how the factors confound each other vary depending on cancer types (Table 14) [73, 76]. At present, the descriptions of related biomarkers in the indications of PD-1/PD-L1 inhibitors are only as follows: “pembrolizumab for advanced/recurrent non-small cell lung cancers [It may be use monotherapy if tumor tests positive for PD-L1. In regard to the PD-L1 expression ratio of tumor cells (Tumor Proportion Score; TPS), become familiar with the “related clinical trials”. It should be tested by pathologists with sufficient experience, in examination facilities, and using vitro diagnostic development.],” and “pembrolizumab for advanced/recurrent MSI-H solid tumors that progressed following cancer chemotherapy.” However, it is very likely that indications based on each biomarker will increase as clinical studies progress and new findings are obtained in the future. Because there was no correlation between the presence of PD-L1 expression and the therapeutic effect of nivolumab in patients diagnosed with dMMR in the Checkmate-142 study [60], PD-1/PD-L1 inhibitors are expected to be effective even when the tumor is negative for PD-L1 expression, as long as it is dMMR.

Thus, at present, TMB or PD-1/PD-L1 testing is not essential to determine the applicability of PD-1/PD-L1 inhibitors. However, it is very likely that they will be recommended in the future to further select patients for whom PD-1/PD-L1 inhibitors are expected to be effective.

Requirements for the quality assurance of testing need to be considered in terms of facility certification, test details, levels and qualifications of testers, staff education, and risk management. It is desirable that testing facilities ensure the reliability of the precision of testing by obtaining and maintaining ISO 15189 (medical laboratories—requirements for quality and competence), an international standard, or external certifications by the College of American Pathologists (CAP) or other organizations. The quality assurance of test details and testers should be implemented according to the “OECD Guidelines for Quality Assurance in Molecular Genetic Testing,” “Japanese Best Practice Guidelines for Genetic Testing, Commentated Edition,” or other relevant documents. For the handling of specimens, please refer to the “Guidelines on the Handling of Pathological Tissue Samples for Genomic Medicine.”

Deficient DNA mismatch repair testing, which is used for determining eligibility for PD-1/PD-L1 inhibitors, has been utilized for screening or as an auxiliary diagnostic method for Lynch syndrome. Therefore, when dMMR testing is performed, informed consent should be obtained after explaining that this test can also be used as screening for Lynch syndrome (refer to “Clinical Practice Resources of the Japanese Society for Familial Tumors” and “Guidance on Genetic Testing in the Clinical Practice of Colorectal Cancer, Third Edition, edited by the Japanese Society of Medical Oncology, November 2016”). As part of the basic clinical practice of cancer, it is assumed that a patient’s family history is taken at the first visit. However, if the patient has been found to have dMMR, the possibility of Lynch syndrome should be reevaluated by checking his or her family history again or through other methods. On the assumption that genetic testing may be considered, a system to provide expert consultation and genetic counseling about the interpretation of test results, subsequent healthcare, heredity in relatives, and other relevant topics must be established in the institution or partner institutions.

Please refer to the following e-learning sites created as part of the construction of a nationwide unified genetic analysis/diagnostic system and the development of a training program by an expert panel consisting of experts from multiple institutions and multiple occupations in collaboration with related academic societies such as the Japanese Society of Medical Oncology:

Immune checkpoint inhibitors activate and maintain tumor immunity by blocking co-inhibitory molecules, which work to suppress immunity in various immune cells. Unlike conventional cytotoxic anticancer drugs or molecular targeted drugs, they do not act directly on cancer cells. They exert their effect by activating immune cells. Since irAEs may occur due to the activation of immune cells, whole-body management is required. Because a delay in response and treatment can lead to a fatal course, immune checkpoint inhibitors should be administered in an environment, where adequate measures can be taken (for the handling of each adverse event, refer to the “Guidelines for Cancer Immunotherapy”, and for measures in each cancer type, refer to the “Optimal Use Promotion Guidelines” in addition to the “Guidelines for Cancer Immunotherapy”).

It is recommended to meet the following criteria (excerpted from “Optimal Use Promotion Guidelines”):

(1) About institutions

Designated cancer hospitals and advanced treatment hospitals designated by the Minister of Health, Labour and Welfare, designated cancer hospitals designated by the prefectural governor, and other hospitals that have physicians with sufficient experience in cancer treatment including cancer pharmacotherapy.

(2) About a system to manage pharmaceutical information within the hospital

The hospital has full-time staff engaged in the pharmaceutical information management and an established system to promptly implement the following actions: liaison for receiving information from pharmaceutical companies; management of pharmaceutical information including that on efficacy and safety, and provision of information to physicians; reporting of any adverse events if they should occur; and others.

(3) About the handling of adverse drug reactions

The hospital has an established 24-h clinical care system that can promptly provide proper diagnosis and treatment of adverse drug reactions in case serious adverse drug reactions occur. Because there are a variety of irAEs, a system to cooperate with experts specializing in respective organs and pathologies needs to be established at the institution or partner institutions. Moreover, it is desirable to have an established team medical care system in which healthcare professionals who are engaged in cancer-related clinical practice and have specialized knowledge and skills perform screening for pain, including monitoring for adverse drug reactions, and share the information with attending physicians.

The approval status in Japan and by the FDA are shown in Supplemental Tables S3 and 4.