Introduction: Invasive micropapillary carcinoma (IMPC) of the breast is a rare and aggressive subtype of invasive ductal carcinomas, associated with poor prognosis and without a well-established treatment. Programmed death-ligand 1 (PD-L1) expression, high tumor-infiltrating lymphocytes (TILs), and microsatellite instability have recently been linked to susceptibility to immunotherapies against PD-1/PD-L1 axis. No exhaustive data is available on the status of these predictive markers in IMPCs of the breast. The aim of our study is to analyze PD-L1 expression, stromal TIL (sTIL), and mismatch repair (MMR) gene status in IMPCs of the breast, to extend the therapeutic possibilities of these rare aggressive tumors. Materials and Methods: Thirty-seven cases of IMPCs diagnosed in two European institutions between 2003 and 2017 with detailed clinical and pathologic data were analyzed. sTILs were assessed in hematoxylin and eosin-stained sections. MMR deficiency was tested by either immunohistochemistry (IHC) for MMR proteins (MLH1, MSH2, MSH6, and PMS2) or capillary electrophoresis for microsatellite instability using a standardized panel of five loci (Bat25, Bat26, D2S123, D5S346, and D17S250). For PD-L1, expression in both tumor cells (TCs) and immune cells (ICs) was determined using the antibody clone SP263. Results: The median sTILs was 3% (mean: 6%, range: 0–40). Thirty-one cases (84%) showed ≤10% of sTILs and only one case had 40% of sTILs. Higher median TILs were more frequently observed in lymph node metastases. PD-L1 expression (≥1%) was observed in 4 (11%) and 14 (38%) cases in TCs and ICs, respectively. None of the tumors showed PD-L1 expression in >1% of TCs. Only three cases showed expression in >10% of ICs. All cases were microsatellite stable by either IHC or polymerase chain reaction analyses. Conclusions: IMPCs of the breast are microsatellite-stable and immune desert tumors with low PD-L1 expression, thus arguing against the use of immune-checkpoint inhibitors in these patients. Active immunotherapy strategies attempting to stimulate self-immune system to attack tumor are needed.
Invasive micropapillary carcinoma (IMPC) of the breast represents a rare (0.7%–3%) and aggressive subtype of invasive ductal carcinoma (IDC) and is associated with poor prognosis and high incidence of lymph node metastases (LNM). This tumor was first described in 1993 as a new entity with characteristic morphological features.[2,3] Histologically, IMPCs show a peculiar architecture, characterized by pseudopapillary structures, composed of groups of cells with inverted polarity, that float in empty spaces and are lined by a delicate layer of fibrous stroma.[4,5] The micropapillary histological pattern has also been found in other organs, such as urinary bladder, colon, lung, or salivary gland, and in all of them, it is associated with poor clinical prognosis and high propensity for lymphatic dissemination.[6,7] The basis of the biological behavior of IMPCs is not yet clear and no specific molecular alterations have yet been identified as therapeutic targets, thus limiting therapeutic options.
Recently, immunotherapy has been introduced in clinical practice for the treatment of several neoplasms, achieving good therapeutic responses, with reduced collateral effects for patients. Immune-checkpoint inhibitors (ICIs) such as pembrolizumab, nivolumab, or ipilimumab are particularly effective in melanoma and metastatic non-small cell lung cancer.[8,9] Positive results have also been found in colon, urinary bladder, or squamous cell carcinoma of the head and neck and others.[10–12] In breast cancer (BC), initially considered a nonimmunogenic tumor, the US Food and Drug Administration has recently granted accelerated approval to atezolizumab in combination with chemotherapy for the treatment of unresectable locally advanced or metastatic triple-negative (TN) BC (TNBC).
Many efforts are currently directed at identifying appropriate biomarkers and immune profiles that can be used to predict immunotherapy responses. Deficiency in the DNA mismatch repair (MMR) system, high tumor mutation burden, programmed death-ligand 1 (PD-L1) levels, and other features of the tumor microenvironment are among the most studied biomarkers. In BC, the presence of tumor-infiltrating lymphocytes (TILs) and PD-L1 expression correlates with the prognosis and antitumor activity of targeted therapies and immunotherapies in HER2+ and TN subtypes.[15,16]
To our knowledge, no study has specifically investigated immune-related biomarkers in IMPCs. In order to understand if ICIs may represent a new therapeutic opportunity, this study aimed to analyze the status of MMR system, PD-L1, and TILs in IMPCs of the breast.
Materials and Methods
Thirty-seven cases of IMPCs were selected from the archives of two hospitals (Department of Pathology, University General Hospital of Catalonia, Barcelona, Spain and Pathology Section, Department of Advanced Biomedical Sciences, University of Naples Federico II, Naples, Italy), from 2003 to 2017. The histological features of all cases were reviewed by three expert pathologists (ND, AE, and SS) by hematoxylin and eosin (H and E) staining. Hormone receptor (HR) and HER2 status information was determined according to the American Society of Clinical Oncology/College of American Pathologists guidelines.[17–19] HER2 equivocal (2+) cases by immunohistochemistry (IHC) were verified by fluorescence in situ hybridization. Clinical characteristics of the patients were retrieved from the hospitals' health information systems. The study was approved by the institutions' review boards.
Stromal tumor-infiltrating lymphocyte assessment
Stromal TILs (sTILs) were semi-quantified using H and E-stained slides. The percentage of sTILs was determined as the area of stromal tissue (i.e., area occupied by mononuclear inflammatory cells over total stromal area) following the methodology recommended by the International TILs Working Group in Breast Cancer.
Programmed death-ligand 1 immunohistochemistry
IHC for PD-L1 was performed in formalin-fixed and paraffin-embedded (FFPE) tissue samples, following manufacturer's instruction. Briefly, 4 μm-cut sections were stained using the VENTANA Benchmark ULTRA autostainer (Ventana Medical Systems, Tucson, AZ, USA). Slides were incubated with a prediluted rabbit monoclonal primary antibody against PD-L1 (clone SP263, #790-4905, Ventana Medical Systems, Tucson, AZ, USA). The reaction was detected using the OptiView DAB IHC detection kit (#760-700, Ventana Medical Systems, Tucson, AZ, USA). A placental tissue (positive) and a normal tonsil (which contains negative, weekly positive, and strongly positive cells for PD-L1) were used as controls. PD-L1 evaluation in both tumor cells (TCs) and immune cells (ICs) accompanying the stroma around the tumor was performed. A trained pathologist evaluated the percentage of TCs with PD-L1 positivity at any intensity above background staining. PD-L1 expression levels were reported in five scoring subgroups: 0 = <1%; 1 = 1%–5%; 2 = >5%–10%; 3 = >10%–24%; 4 = ≥25%–49%; and 5 = ≥50%. The cases showing membrane positivity (complete or incomplete) in ≥1% of TC or any pattern of positivity (cytoplasm or membrane) in ≥1% of ICs were considered as positive.
Analysis for mismatch repair system deficiency
IHC staining was performed on 4-μm FFPE sections at the University General Hospital of Catalonia, using monoclonal antibodies against MLH1 (clone G168-728, prediluted), MSH2 (clone G219-1129, prediluted), and MSH6 (clone 44, prediluted) from Ventana Medical Systems (Tucson, AZ, USA) and PMS2 (clone A16-4, dilution 1:25; BD Biosciences, San Jose, CA, USA). The staining patterns of MMR proteins were evaluated using stromal or inflammatory cells as internal controls. Absence of nuclear reactivity of TCs was considered as MMR deficiency (dMMR). In 26 cases, tissue samples were available for the evaluation of microsatellite instability (MSI) by polymerase chain reaction (PCR), at the laboratory of the Department of Public Health, University Federico II, Naples, using a primer set sensitized basing on Bethesda consensus without the fluorescent label. Briefly, DNA, from tumor specimen and its corresponding normal tissue, was extracted from FFPE tissues. Multiplex 3-marker PCR microsatellite analysis was carried out as previously described. For each patient, extracted DNA, from tumor and paired normal tissue, was amplified with specific primers for BAT25, BAT26, D2S123, D5S346, and D17S250 by using AmpliTaqGold DNA polymerase (Applied Biosystems, Foster City, CA, USA). One microliter of the amplified PCR product was evaluated on TapStation4200 platform (Agilent Technologies, Santa Clara, CA, USA), and the results were analyzed by using 4200 TapeStation software (version 3.1, Agilent, Santa Clara, CA, USA). Following the Bethesda guidelines, a tumor was classified as MSI-high (MSI-H) if two or more markers showed instability, MSI-low if a single marker revealed instability, and MS-stable.
The clinicopathologic characteristics of the patients with IMPCs were compared using the t-test for continuous variables or χ2 tests (or Fisher's exact test when cell frequencies were below 5) for categorical variables. For all analyses, P < 0.05 was considered statistically significant. Statistical analysis was carried out with SPSS software (version 24, IBM, New York, USA).
A total of 37 patients were analyzed in our study. The patients' age ranged from 29 to 83 years. The tumors were located on the right breast in 20 cases (54%) and on the left side in 17 cases (46%). Twenty-eight (76%) cases showed a pure micropapillary pattern. Sixteen (43%) cases were of histologic grade 3 and the remaining cases were of grade 2 tumors. Twenty-six (70%) patients had T1 tumors, 6 (16%) showed T2 tumors, and 5 (14%) had T3 tumors. All cases had axillary LNM. Twenty-six cases (70%) were included in the HR+ subtype, 10 cases (27%) were HER2+, and 1 case (3%) was a TNBC. The samples analyzed were from primary tumor (n = 17, 46%) and LNM (n = 20, 54%). The clinicopathological features of the patients are reported in Table 1.
Stromal tumor-infiltrating lymphocytes
The median sTILs was 3% (mean: 6%, range: 0–40). Absence of sTILs was observed in eight cases (22%). Twenty-three (62%) and six (16%) cases showed between 1%–10% and >10% sTILs, respectively. sTILs greater than the median value were more frequently observed in LNM (Pearson's Chi-square, P = 0.028). Only one case (a LNM) had 40% of sTILs. No significant differences between sTIL levels, HR status, and HER2 status were found [Table 2 and Figure 1].
Programmed death-ligand 1 expression
PD-L1 expression was analyzed in both TCs and ICs [Figure 1]. Positive TC staining (≥1% of cells) was found in four cases (11%). The percentage of stained TCs was never above the 1% positivity threshold. In 14 cases, we found the expression of PD-L1 in at least 1% of ICs. Only three cases showed expression in >10% of ICs [Table 3]. PD-L1 TC positivity was more frequent in HER2+ as compared to the other subtypes (30% HER2+ versus 4% HR+/TNBC, Pearson's Chi-square, P = 0.05). No associations between PD-L1 expression and other clinicopathological features were found [Table 4].
Correlation between stromal tumor-infiltrating lymphocytes and programmed death-ligand 1 expression
High sTILs (>10%) were significantly associated with PD-L1 positivity (≥1%) in both TCs (Fisher's exact test, P = 0.0096) and ICs (Fisher's exact test, P = 0.0001). No associations were found when using the median sTIL value as the cutoff point (TCs, P = 0.31; ICs, P = 0.33) [Table 4].
Analysis of the DNA mismatch repair system
Loss of MMR system was studied in 37 cases with IHC technique and in 26 cases with capillary electrophoresis (CE) technique. All tumors were classified as stable because neither loss of protein expression nor alterations of the foci in the molecular study were identified.
Immunotherapy in cancer treatment has shown many progresses in recent years and represents an effective therapeutic tool for various tumor types.[9,8] In BC, the efficacy of PD1/PD-L1 pathway inhibitors has been demonstrated in TN subtype[13,24,25] where predictive biomarkers of response to ICIs[26,27] such as TILs and PD-L1 are found expressed at higher levels as compared to other BC subtypes.[28–30] However, limited data are available on the status of these biomarkers in rarest BC subtypes. IMPCs are considered a special category of breast epithelial cancer, representing 0.7%–3% of all BCs, that generally show a luminal phenotype (HR+ with or without HER2 amplification), associated with poor outcome and limited therapeutic options.
To the best of our knowledge, this is the largest set of BCs with micropapillary pattern in which PD-L1 expression has been analyzed. In a previous study across BC subtypes, no PD-L1 expression was observed in two cases of IMPC analyzed. We found that 38% of IMPCs expressed PD-L1 in ICs (≥1%), similar to the positivity rate described in luminal B tumors (35%). This percentage is significantly lower than what has been reported in TNBC. A recent biomarker analysis in patients with metastatic TNBC (mTNBC) treated with single-agent atezolizumab, found PD-L1 expression (≥1% in ICs) in 78% of patients. High levels of ICs (IC3, >10%) were independently associated with higher overall response rate and overall survival in mTNBC. In our study, only three IMPC cases (8%) showed PD-L1 expression in ICs >10%, being all from LNM. The higher expression of PD-L1 in LNM as compared to primary tumors has been already described in BC. One hypothesis that could explain this difference is that the enriched infiltrating T cells in lymph nodes may drive PD-L1 expression to induce adaptive immune resistance during infiltration of TCs. If this is true, LNM could represent a better surrogate for determining PD-L1 expression in optimal treatment selection. Second, preexisting PD-L1 expression in T cells and macrophages in the lymph node microenvironment may be tumor unrelated. Further studies are needed to confirm if this difference has an impact on response to anti-PD-L1/anti-PD-1 therapy. We found that 11% of IMPCs expressed PD-L1 in TC (≥1%). Interestingly, three out of four TC-positive tumors were HER2+. These tumors showed PD-L1 ICs ≥5% and an average TIL of 18%, thus suggesting that IMPCs of this subtype behave similar to the conventional HER2+ BC.
The presence of TILs is a favorable prognostic factor in BC, and TILs may synergize with chemotherapy and ICI therapy for improved clinical response.[35,36] A median of 11% (range, 5%–26%) of BC are lymphocyte predominant BC (LPBC) with higher incidence in TNBC (20%) followed by HER2+ (16%) and HR+/HER2− (6%). In our study, IMPCs exhibited a predominant luminal phenotype (72% of HR+/HER2−) and poor IC infiltrate, with only 6/37 cases (16%) showing >10% of sTILs. None of the IMPCs analyzed was an LPBC, suggesting that IMPCs may share the same immunosuppressive microenvironment with the less aggressive luminal BC. Our results are in contrast with a previous finding, reporting a higher rate of lymphocytic infiltration in IMPCs (55%, 28/51). The same authors found that prominent lymphocytic infiltration was associated with LNM and that lymphocytes were predominantly CD4+ and confined almost exclusively in the tumor stroma.[38,39] This difference may be explained by the different representation of IMPCs with pure micropapillary pattern in the two studies (76% vs. 7%, 4/51). In addition, Guo et al. did not report the HR and HER2 status of their tumors and used a different evaluation methodology for lymphocyte infiltration (qualitative rather than semi-quantitative as in our study) that limit any comparative analysis. In our study, we could not associate the presence of LNM with sTIL abundance as all patients included were node positive.
MSI increases susceptibility to ICIs of some tumor types. As a matter of fact, in 2017, pembrolizumab was granted accelerated approval for treatment of adult and pediatric patients with unresectable or metastatic MSI-H or dMMR solid tumors, progressing following prior treatment and who have no satisfactory alternative treatment options, representing the “first tissue/site-agnostic approval.”[40,41] In BC, low prevalence or absence of dMMR (<1%) has been observed in HER2+ and TNBC IDCs and mucinous carcinomas. In our study, we confirmed the absence of dMMR in the IMPCs, as observed in other organs, where dMMR associated to a micropapillary pattern seems to be a very uncommon event.
Our study has some limitations. Despite being the largest study investigating the status of predictive biomarkers of ICI response in IMPCs, our patients' cohort is still small and studies with more cases including patients which did not metastasize are needed. Second, we did not have outcome data of the patients included in our cohort. This would have allowed additional correlative analyses enhancing the clinical significance of our work. Third, sTILs were assessed using the standard H and E evaluation. We neither performed in-depth phenotyping nor studied the activity of the ICs populating IMPC microenvironment. Lastly, our PD-L1 expression data rely on the application of a single assay (clone SP263). Therefore, results may be varied upon different PD-L1 antibodies and scoring system used.
Our data suggest that IMPCs are microsatellite-stable tumors with an immunosuppressive microenvironment and low PD-L1 expression. The potential of IMPCs to evade the immune system may be supported by multiple mechanisms. Molecules involved in natural killer T cells (CD1d), macrophages (PJA2), and cytotoxic T cells (Granzyme A precursor) antitumor activity have been shown to be expressed at lower levels in IMPC than in conventional IDC of the breast. Active immunotherapy strategies attempting to stimulate self-immune system to attack tumor are, therefore, needed.
Financial support and sponsorship
The authors disclosed no funding related to this article.
Conflicts of interest
PN has consulted for Bayer, Novartis, and MSD and received compensation. The other authors declare no potential conflict of interest.
*These authors contributed equally as co-senior authors
For reprints contact:email@example.com