The approval of pembrolizumab for treatment of patients with microsatellite instability-high (MSI-H) or mismatch repair–deficient (dMMR) advanced cancers has led to increased requests for MSI and/or MMR immunoperoxidase (IPOX) testing. Diagnoses for patients with advanced-stage cancer are frequently made from cytology specimens.
To investigate the feasibility of using cell block (CB) preparations of effusions for MMR IPOX evaluation.
Surgical pathology cases of colorectal and endometrial carcinomas with known MMR/MSI status and matched effusions with available CBs were identified. Cell block sections were evaluated for adequacy and stained with MMR IPOX (MSH2, MSH6, MLH1, and PMS2). The CBs were reviewed, the number of tumor cells quantified, and MMR IPOX was interpreted as retained, lost, suboptimal, or noncontributory.
We identified 748 cases with MMR/MSI testing on surgical specimens having matched effusions. Of these, 131 cases (17.5%) had an available CB and 53 were deemed adequate for MMR IPOX staining. MMR IPOX results between effusion CBs and surgical pathology specimens were concordant in 45 of 53 (85%), inconclusive in 6 of 53 (11%), and discordant in 2 of 53 (4%) cases.
There was high concordance of MMR IPOX testing between cytologic and surgical specimens, with no false-positive and 2 false-negative CB results. Limited tumor cells, staining in cells indefinite as tumor, tumor staining heterogeneity, and lack of internal control staining were problematic in some cases. Our findings indicate that cytologic effusion specimens may be suitable substrates for MMR IPOX biomarker testing; however, inconclusive cases need to be interpreted with caution.
DNA mismatch repair (MMR) is a highly conserved process involving the 4 key genes mutL homologue 1 (MLH1), mutS homologue 2 (MSH2), mutS homologue 6 (MSH6), and postmeiotic segregation increased 2 (PMS2), which act to identify and repair mismatched base pairs that may arise during DNA replication.1 This system can develop functional mutations by various means, especially in regions of repetitive nucleotide sequences called microsatellites, resulting in an alteration referred to as microsatellite instability (MSI).
Recent studies have shown that patients with microsatellite instability-high (MSI-H) or mismatch repair–deficient (dMMR) advanced cancers can benefit from treatment with the immune checkpoint inhibitor pembrolizumab. Dudley et al2 and Le et al3 hypothesized that dMMR tumors have an increased mutational burden that creates neoantigens that can be detected by the native immune system.4–6 Therefore, enhancing the immune response may be beneficial. This could be achieved by blocking the programmed death receptor-1 (PD-1) pathway through administration of the anti–PD-1 antibody pembrolizumab, thus releasing the negative feedback system on the Th1 cytotoxic immune response. In the Le et al study, patients with dMMR solid tumors (colorectal and noncolorectal types) had improved immune-related objective response rates and immune-related progression-free survival rates as compared to patients with MMR-intact colorectal cancers treated with pembrolizumab.3
MMR status is routinely assessed in colorectal and endometrial carcinoma as a method of cancer prevention, surveillance in patients with Lynch syndrome and their families, and for prognostic, predictive, and therapeutic implications.7–11 In 2017, the United States Food & Drug Administration (FDA) granted accelerated approval of pembrolizumab to treat pediatric and adult patients with unresectable or metastatic MSI-H or dMMR solid tumors that have progressed despite prior therapy.12 As a result, there have been an increased number of requests from clinicians for MMR immunoperoxidase (IPOX) and/or MSI polymerase chain reaction (PCR)–based tests in this patient population. MMR IPOX and MSI PCR-based testing is typically performed on surgical pathology (SP) biopsy or resection specimens. However, patients with unresectable and/or metastatic disease often present at an advanced stage with body cavity effusions, and those cytologic samples may be the only diagnostic material available for MMR evaluation.
In the pathology era of “less is more,” cytology specimens have become increasingly used for diagnoses and ancillary studies. Preparation of a cell block (CB) in cytology is a common specimen-processing method, frequently akin to that of an SP tissue block, and can be advantageous in IPOX staining and molecular and biomarker testing.13 While most CBs are processed similarly to SP formalin-fixed, paraffin-embedded tissue, there are a variety of preanalytic factors in cytology, such as different collection media, preservatives, and fixative solutions, which may impact protein antigenicity, resulting in altered or false-negative IPOX staining.14
For these reasons, we endeavored to assess the feasibility of using CBs prepared from effusion specimens for MMR IPOX testing in a cohort of patients with colorectal and endometrial carcinomas with known MMR/MSI results from SP specimens. To our knowledge there have been no prior publications comparing the efficacy of MMR IPOX testing on cytologic specimens with SP specimens.
MATERIALS AND METHODS
Following approval from the Institutional Review Board, a retrospective search of our electronic medical record database from January 2000 to November 2017 was performed in order to identify SP biopsies and resections of colorectal and endometrial carcinomas with known MMR/MSI status and having matched effusions with available CBs from the same patient. The hematoxylin-eosin (H&E)–stained CB sections were reviewed for adequacy, defined as containing both sufficient tumor cells and normal cells (lymphocytes, macrophages, mesothelial cells, glandular cells). Tumor cell sufficiency was defined as any number of identifiable tumor cells in the H&E- and MMR IPOX-stained CB sections. The initially examined H&E-stained CB sections were either preexisting or cut from the CB at the time of the study. Cases with insufficient tumor cells on initial H&E sections or on any of the subsequent MMR IPOX-stained sections were excluded.
Cell Block Preparation
Cytologic effusion specimens were collected at our institution by using standard techniques, received fresh in the cytology laboratory, and processed within 48 hours. A CB was prepared according to previously described techniques and our laboratory protocol.15 Per protocol, for effusion cell block preparations, initially 250 mL of the effusion fluid (divided into five 50-mL tubes) was centrifuged at 377g for 10 minutes. After decanting the supernatant, sediment cell pellets from each tube were combined and 4 cytospin slides prepared (1 Diff Quik and 3 Papanicolaou stained) by using 1 to 2 drops of the combined pellet. Cell blocks were made from the combined pellet after adding 5 mL of 95% ethanol and 5 mL of 10% formalin, centrifuging this material at 377g for 10 minutes, and decanting the supernatant. The resultant cell pellet was then removed with a spatula, wrapped in filter paper (Shark Skin filter paper, GE Healthcare Whatman), placed into a labeled tissue cassette, and immersed in 10% formalin for fixation and routine histologic processing. After paraffin embedding, 4-μm-thick section(s) were cut and stained with H&E or left unstained for other ancillary tests.
Immunoperoxidase staining for MLH1 (1:300; clone G168-728, Cell Marque), PMS2 (1:125; clone A16-4, BD Biosciences), MSH2 (1:100; clone FE11, Calbiochem), and MSH6 (1:300; clone 44, BD Biosciences) was performed on unstained CB sections with the Leica Bond III stainer (Rankin Biomedical Corporation). MMR IPOX was interpreted as retained, lost, suboptimal, or noncontributory (Figure 1, A through E; and Figure 2, A through T). Retained staining was defined as any perceivable nuclear staining in any percentage of tumor cells. The intensity of positive staining in tumor cells was compared to the intensity of staining in internal control cells. Loss of staining was defined as complete absence of nuclear staining in all tumor cells but retained in normal cells. Suboptimal staining was defined as questionable staining of definitive tumor cells (S-QS) or focal staining of cells that were indefinite for tumor (S-CIFT). Noncontributory (NC) staining was used to describe those cases where there was lack of staining in both internal control cells (ie, normal cells such as lymphocytes, mesothelial cells, and glandular cells) and tumor cells.
H&E-stained CB sections were reviewed to determine adequacy for MMR IPOX staining as defined above. For each CB that was deemed adequate, the number of tumor cells in the CB was visually estimated and categorized as being fewer than 10, 10 to 50, 51 to 300, and more than 300 cells. Tumor was identified largely on the basis of cytomorphologic features. In some cases, confirmatory IPOX testing (ie, MOC31, Ber-EP4, CDX2) was previously performed as part of the diagnostic workup. In such cases, the confirmatory IPOX slides were compared with the subsequent H&E- and MMR IPOX-stained CB sections in order to correctly map out tumor cell locations. Three pathologists, who were blinded to the corresponding SP MMR IPOX or MSI-PCR test results, reviewed the cases and interpreted the MMR IPOX as retained, lost, suboptimal (S-QS or S-CIFT), or NC. Concordant cases were those in which the MMR IPOX staining of the 4 MMR proteins (MLH1, PMS2, MSH2, MSH6) of cytology CBs showed the same result as MMR IPOX staining and/or MSI PCR testing of SP specimens from the same patient. Conversely, discordant cases were those in which the MMR IPOX staining of the 4 MMR proteins (MLH1, PMS2, MSH2, MSH6) of cytology CBs were incongruous with the MMR IPOX staining and/or MSI PCR testing of SP specimens from the same patient. Inconclusive cases were those cases in which MMR status could not be determined owing to problematic IPOX staining of the individual MMR proteins (ie, S-QS, S-CIFT, and NC, as defined above).
We identified 748 patients from our database, diagnosed with colorectal or endometrial carcinoma with MMR IPOX and/or MSI PCR testing on SP specimens, who had a matched effusion specimen. Of the effusion specimens, 131 cases (17.5%) from 107 patients had an available CB. After review of the 131 H&E-stained CB sections, a total of 53 CBs (40.5%) from different effusion specimens in 47 patients were deemed adequate for MMR IPOX testing. These included 30 ascitic, 21 pleural, and 2 pericardial effusion specimens (Figure 3).
The median patient age was 55 years (range, 24–83 years) with slightly more females than males (61.7% [29 of 47] and 38.3% [18 of 47], respectively). Of the 53 cases, there were 7 (13%) of endometrial carcinoma and 46 (87%) of colorectal carcinoma. Nine of 47 patients (19.1%) received no chemotherapeutic agents, 2 of 47 (4.3%) were treated with 1 chemotherapeutic agent, and 36 of 47 (76.6%) were treated with 2 or more chemotherapeutic agents before the effusion specimen collection. Patient demographic characteristics and clinicopathologic features are summarized in Table 1.
Upon review, the MMR IPOX results between effusion CB and SP specimens were concordant in 45 of 53 (85%), inconclusive in 6 of 53 (11%), and discordant in 2 of 53 (4%) cases. In the concordant cases (1 of which is illustrated in Figure 1, A through E), 2 of 45 (4%) had fewer than 10 tumor cells (1 of which is illustrated in the supplemental digital content at https://meridian.allenpress.com/aplm in the January 2021 table of contents), 8 of 45 (18%) had 10 to 50 tumor cells, 17 of 45 (38%) had 51 to 300 tumor cells, and 18 of 45 (40%) had more than 300 tumor cells in the CB (Figure 4; Table 2). Concordant cases included 39 of 45 colorectal carcinomas (87%) and 6 of 45 endometrial carcinomas (13%), of which 25 of 45 (56%) were from ascitic fluids, 18 of 45 (40%) were from pleural fluids, and 2 of 45 (4%) were from pericardial fluids.
Of the 6 inconclusive cases (4 of which are illustrated in Figure 2, A through T), 5 of 6 (83%) were colorectal carcinomas, 1 of 6 (17%) was endometrial carcinoma, 3 of 6 (50%) were from ascitic fluids, and 3 of 6 (50%) were from pleural fluids. Half of the cases (3 of 6, 50%) had 10 to 50 tumor cells and the remaining half (3 of 6, 50%) had 51 to 300 tumor cells in the CB (Figure 4; Table 2). Half of the patients did not receive treatment before the effusion specimen collection and the other half were treated with 2 chemotherapeutic agents before specimen collection.
Both discordant cases were from ascitic fluid specimens in patients with colorectal carcinoma and had 51 to 300 tumor cells in the CB (Figure 4). One patient did not receive treatment before effusion specimen collection and the second patient had prior treatment with 2 chemotherapeutic agents.
We also attempted to assess if any specific MMR protein might be responsible for the inconclusive and discordant results. In the inconclusive cases, there was difficulty in interpreting MSH6 in 5 of 6 cases (83%), MLH1 in 1 of 6 cases (17%), and MSH2 in 1 of 6 cases (17%). In the 2 discordant cases, 1 case (50%) involved interpretation of MSH6 and the other case involved interpretation of both MLH1 and PMS2 (Figure 5, A). Cases were inconclusive owing to S-CIFT in 4 of 6 (67%), NC staining in 2 of 6 (33%), and S-QS in 1 of 6 (17%) cases (Figure 5, B). In inconclusive cases, S-CIFT was reported in the interpretation of MLH1 in 1 of 6 (17%), MSH2 in 1 of 6 (17%), and MSH6 in 2 of 6 cases (33%); NC was reported in interpretation of MSH6 in 2 of 6 cases (33%); and S-QS was reported in the interpretation of MSH6 in 1 of 6 (17%) cases (Figure 5, C). However, the sample size for inconclusive/discordant cases was too small for meaningful statistical analysis.
The most common cause for an inconclusive result was suboptimal interpretation due to focal staining in cells indefinite for tumor (S-CIFT), which was recorded in 4 of 6 inconclusive cases (67%) (Figure 5, B). However, it is important to note that all inconclusive cases had at least 2 MMR proteins that were able to be interpreted and 5 of 6 (83%) had 3 markers that were interpretable.
With the FDA approval of the use of the immune checkpoint inhibitor pembrolizumab for patients with MSI-H or dMMR advanced solid cancers, there have been and will continue to be increased requests from clinicians for MMR IPOX and/or MSI PCR testing of patient specimens. Since cytology samples, such as effusion specimens, may be the only material available for evaluation in patients with advanced cancers, in this study we evaluated the feasibility of determining MMR status on CB preparations of these specimens by using MMR IPOX staining.
In our study, there was high concordance (85%) of MMR IPOX results between cytology effusion CB specimens and SP specimens, with the CB results showing no falsely retained staining and only 2 falsely lost staining CB results. Overall, MMR IPOX was easily interpretable, even in CBs with very limited (fewer than 10) tumor cells or with a predominant single cell pattern.
Interpretation of MMR IPOX requires the presence of staining in internal control cells (ie, normal cells such as lymphocytes, mesothelial cells, and glandular cells). Appropriate staining of internal control cells was problematic in some of our cases, which is especially important for interpretation of those cases with loss of MMR protein expression. Repeated IPOX staining or MSI PCR testing in such instances may be considered.
Of the MMR proteins, MSH6 most commonly caused issues with interpretation. This finding is not entirely unexpected, as MSH6 is known to show heterogeneous and weak staining in tumors.16 All inconclusive cases had at least 2 MMR proteins that were interpretable, 5 of 6 (83%) had interpretable results for PMS2 and MSH2, 5 of 6 (83%) had interpretable results for MLH1, and 1 of 6 (17%) had interpretable results for MSH6. Some studies have shown that in samples with limited tissue, MMR proteins MSH6 and PMS2 may be used as surrogate markers for MSH2 and MLH1, respectively. MSH6 and PMS2 are obligate binding partners of MSH2 and MLH1, respectively, and in the absence of their binding partners (MSH2 and MLH1) they are not expressed at the protein level. Thus, loss of MSH6 expression can detect both defects in MSH6 and MSH2, whereas loss of PMS2 expression is a proxy for both PMS2 and MLH1 defects.11,17,18 Also, if an effusion specimen shows retention of MLH1, then it can be inferred that there is no MLH1 promoter methylation.19 Therefore, even though the overall MMR status may be inconclusive, valuable information can still be gained.
Various preanalytic factors, including type of fixative used, time in formalin before embedding, and uniformity of fixation, have been cited as being confounding factors in interpretation of MMR IPOX staining. However, it has also been shown that there is more uniform and complete cellular fixation in biopsy samples than with SP resection specimens, and some publications have even indicated that MMR IPOX testing on biopsy samples may be preferable to testing on resection samples.20–22 Cell block specimens are similar to biopsy specimens in that they are small and will likely have more complete and uniform fixation as they are processed. Although the processing specifics of each specimen cannot be fully investigated in this retrospective study, it is unlikely that these preanalytic variables were contributory to our inconclusive and discordant results, as all cases were processed in our laboratory and in the same manner (please refer to Materials and Methods section). Other challenging aspects of working with CBs include inconsistent and/or suboptimal cellularity, multiple preparatory techniques using a variety of fixatives, inability to perform on-site adequacy assessment, and appropriate specimen triage, among others.23
Neoadjuvant chemotherapy has also been shown to have potential effects on MMR IPOX staining. Some studies showed reduced MMR IPOX staining in patients with colorectal cancer who were undergoing neoadjuvant chemotherapy treatment.24,25 In our study, half of the patients with inconclusive results had no treatment before effusion specimen collection and the other half were previously treated with 2 chemotherapeutic agents, suggesting no definitive correlation between prior treatment and MMR IPOX results.
In general, the establishment of standardized criteria for determining positivity and subsequent eligibility for targeted therapies using biomarker testing platforms is challenging. For example, different cutoff values and specimen adequacy limits for reporting positivity in programmed death ligand-1 (PD-L1) IPOX staining have been proposed depending on the tumor type and antibody clone used.26,27 In MMR IPOX staining, any perceivable nuclear staining is considered positive. However, MMR IPOX staining may be heterogeneous, and a standardized number of tumor cells needed for interpretation in cytologic specimens has not been established.16,28 In an effort to define a minimal number of tumor cells needed for adequate assessment, we semiquantified the number of tumor cells in each case. Number of tumor cells ranged from 7 to more than 300 in concordant cases, 20 to close to 300 in inconclusive cases, and 60 and 170 in the 2 discordant cases. These findings do not reveal any specific cutoff for the number of tumor cells above which results are concordant. Nevertheless, although the sample size is limited, it may be important to note that all cases with more than 300 tumor cells were concordant with SP results.
The most common cause of inconclusive interpretations was cases in which staining occurred in cells that were indefinite for tumor (S-CIFT). In such cases, confirmatory IPOX testing may be considered to verify tumor cells. Both discordant cases were likely due to falsely lost staining in CB sections where MMR IPOX staining was interpreted as lost and the SP specimen showed retention of MMR proteins. Tumor staining heterogeneity, sampling, and possible loss of MMR protein expression in metastases may also have played a role in inconclusive and discordant cases. However, it should be noted that few studies have addressed the concordance of MMR status in primary and corresponding metastatic tumors, with conflicting results, and the loss of MMR protein expression in metastases has not been reliably established.29–31
The main limitation in our study was sample size, as only a small number of CBs were deemed adequate for testing. Furthermore, most cases (51 of 53, 96%) had intact MMR IPOX/MSI PCR on SP specimens and only 2 of 53 cases (4%) showed loss of MSH2 and MSH6 (cases 6 and 49, Table 2). As only 1 of the 2 cases (case 6) with loss of MSH2 and MSH6 showed concordant CB and SP specimen results, more cases need to be studied to confirm these findings. MSI PCR testing, which may be informative in clarifying our inconclusive results, was outside the scope of this study, as we focused on the utility of MMR IPOX in cytology specimens as an inexpensive and easily available option for determining MMR status.
In conclusion, effusion CB specimens may be suitable substrates for MMR IPOX testing in those cases where histologic material is either inadequate or unavailable. However, MMR IPOX staining in CB specimens must be interpreted with caution, and inconclusive cases should be reported with appropriate commentary. Loss of MMR IPOX staining in CB specimens should be confirmed by other means when possible, in particular for MSH6 owing to its known weak and patchy performance even in surgical specimens. Additional studies evaluating the concordance of MMR IPOX and/or MSI PCR testing in cytologic and histologic material are needed, as establishing MMR and/or MSI status is becoming increasingly important in determining patients' eligibility for immunotherapy.
Supported by The University of Texas MD Anderson's Department of Pathology Chairman Funds (EMJ) and NIH SPORE in Uterine Cancer NIH P50 CA09825 (RRB).
Supplemental digital content is available for this article at at https://meridian.allenpress.com/aplm in the January 2021 table of contents.
The authors have no relevant financial interest in the products or companies described in this article.
Our preliminary data were presented as an abstract and in the newly instituted Moderated Poster Session at the 108th Annual United States & Canadian Society of Pathology Meeting; March 18, 2019; National Harbor, Maryland.