Abstract
Background/Aim: Loss of differentiation of breast cancer cells in association with a down-regulated class I human leukocyte antigen (HLA) expression can lead to proliferation unhampered by cytotoxic T lymphocytes, which has been proven to be of prognostic relevance. The objective of this study was to determine the levels of HLA-A and HLA-B/C expression in metastatic breast cancer (MBC) cells and their usefulness for predicting 5-year survival. Materials and Methods: This prospective double-blinded cohort study analyzed patients starting a new line of therapy for MBC. RT-qPCR was used to determine the levels of HLA-A and B/C expression in MBC cells and the mRNA-based tumor intrinsic subtype. Two receiver operating characteristic curves (ROC) were constructed in order to determine whether HLA-A and HLA-B/C expression levels can be used for predicting 5-year survival. Youden J points, and sensitivity and specificity optimized cut-off points were determined for both ROC curves. Results: We enrolled 34 patients. The ROC curve for HLA-B/C had the highest AUC compared to HLA-A (0.55 vs. 0.42). High levels of HLA-A and HLA-B/C expression (40-ΔΔCT of 33.5 and 31.9, respectively) were highly specific (reaching 87.5% for HLA-A and even 100% specificity for HLA-B/C) yet insensitive for five-year survival in our study. Conclusion: High expression of certain class I HLA molecule subtypes by MBCs, in particular high HLA-A or B/C expression by MBC cells seems very specific in predicting the 5-year survival. We determined cut-off values for these HLA molecule clusters with high specificity, which might help identify patients with a favorable prognosis as prognosticators of a 5-year overall survival if their sensitivity is improved in larger prospective cohorts.
Breast cancer (BC) is one of the most prevalent female malignant neoplasms and the most common cause of cancer-related death in women worldwide (1). Approximately 8% of women in Germany are diagnosed with breast cancer at a point when metastatic disease is already present (2). Systemic treatment of patients with metastatic breast cancer (MBC) includes molecularly targeted agents like HER2 antibodies, endocrine therapies, immunotherapies and inhibitors of PIK3CA, PARP, CDK4/6 or mTOR (3). However, the prognosis of MBC is still dismal (4).
It has been well established that tumor growth can be stemmed by cytotoxic (CD8+) T lymphocytes (CTLs) through immune surveillance (5, 6). Namely, tumor antigens can be presented to CTLs in a complex with class I HLA molecules that are normally found on most human cells (7, 8). CTLs activated in such a manner destroy tumor cells through mechanisms that have been described numerous times in the context of immune responses to intracellular antigens (most often viruses and intracellular bacteria) (1, 5, 7).
However, de-differentiation of tumor cells can lead to a reduction or a complete loss of class I HLA expression in such cells (9-11). As one would expect, this phenomenon has been associated with a poor prognosis in several tumors and breast cancer is not an exception (6, 12). Also, tumors with a reduced class I HLA levels tend to be of a higher grade compared to those with normal levels (12). Beta-2 microglobulin (β2m) is paramount for the stability and function of HLA class I molecules and can serve as a surrogate marker for HLA class I expression (13). Finally, some tumor cells have been shown to express placenta-like or immunologically dysfunctional HLA types like HLA-G and HLA-J (14, 15).
The objective of this study was to determine the levels of HLA-A and HLA-B/C expression in MBC cells above which a statistically significant number of patients will survive beyond 5 years.
Materials and Methods
This was a prospective double-blinded cohort study that analyzed 34 patients treated for metastatic breast cancer at the German National Center for Tumor Diseases (NCT), Heidelberg University, Germany, and the Department of Obstetrics and Gynecology, University of Heidelberg, Germany from March 2010 to May 2015. Initiation of a new line of therapy was an enrollment criterion for all patients.
RT-qPCR technology was used in order to determine the extent of HLA-A, B and C expression in breast cancer cells as well as their intrinsic subtype, while utilizing CALM2 as the housekeeping gene, according to established protocols (16).
The relative gene expression was presented as 40-ΔΔCT with higher values corresponding to higher mRNA counts in the tissue biopsy sample.
The study participants were restaged for disease progression every 3 months until loss to follow-up or death. The Response Evaluation Criteria in Solid Tumors (RECIST) system was used for this purpose. Electronic health records and medical histories were also searched in order to precisely define date of death, in case of such an unfortunate outcome.
Two receiver operating characteristic (ROC) curves were constructed in order to determine the cut-off values of HLA-A and HLA-B/C mRNA expression levels in MBC cells that can be useful for predicting survival beyond 5 years. To this end, Youden J points were calculated for the two ROC curves with consecutive determination of sensitivity and specificity at the respective J points. Finally, we identified sensitivity and specificity optimized cut-off points.
Both patients and physicians were unaware of the patients’ MBC HLA-statuses. Those that performed HLA mRNA quantification as well as the radiologists in charge of restaging were blinded to the patients’ medical histories.
The demographic data and clinical characteristics of the patients and the characteristics of their cancer are presented as frequencies and percentages and medians and ranges, as appropriate. ROC curves were analyzed and areas under the curve (AUCs), Youden J points and sensitivities and specificities were defined. All statistics were computed using R (version 3.1.2). A significance level of 5% was chosen.
Results
We enrolled 34 patients into the study. The characteristics of the patients and their MBC are outlined in Table I. The median age at BC diagnosis was 50 years while the median age at study enrollment was 60.5.
Patient and tumor characteristics (n=34).
The most common MBC subtype (38.2%) was the luminal B subtype, while 20.6% had triple negative cancers. Almost a quarter of the patients studied had bone metastases.
The distributions of HLA-A and HLA-B/C expression are presented in Figure 1. The distributions were bimodal in both cases. A Mann-Whitney-U (MWU) test revealed that such distributions were not statistically significantly different when comparing the deceased within 5 years since study enrollment included patients surviving 5 years or longer (MWU=63.5, p=0.51 for HLA-A and MWU=69, p=0.65 for HLA-B/C).
Histograms illustrating human leukocyte antigen (HLA) expression: A) HLA-A, B) HLA-B/C. 40-ΔΔCT: Relative gene expression.
ROC curves graphically representing the diagnostic efficiency of HLA-A and HLA-B/C expression by MBC cells for 5-year survival are displayed in Figure 2. The ROC curve for HLA-B/C has the higher AUC of the two (0.42 vs. 0.55). Youden J points, specificity- and sensitivity-optimized cut-off points are represented in Table II. High levels of HLA-A and HLA-B/C expression (a 40-ΔΔCT of 33.5 and 31.9, respectively) were highly specific (87.5% and 100%, respectively) for five-year survival. The best sensitivity (73.7%) was achieved at a cut-off of 30.5 40-ΔΔCT for HLA-A.
Receiver operating characteristic curves representing the sensitivity and specificity of A) human leukocyte antigen (HLA)-A and B) HLA-B/C expression for the detection of patients likely to survive 5 years post study enrollment (only non-survivors were analyzed).
Receiver operating characteristics.
Discussion
In previous decades, it has been recognized that cellular immunity and CTLs in particular play an important role in the control of neoplastic growth (5, 6). Namely, tumor cell dedifferentiation and concomitant HLA-A, B and C repression lead to an inadequate T-cell response to such tumor cells, since these molecules are paramount for tumor antigen recognition by CTLs (9-11). However, the HLA expression process is not a binary affair and tumor cells, like normal cells, can express varying levels of HLA molecules (17). These concepts have been tested in the context of several neoplasms and BC is one of them. In our study, MBC cells were shown to express relatively high levels of HLA-A and much lower levels of HLA-B or C (Figure 1). Approximately a quarter of our patients lived for 5 years or longer after study enrollment. Sadly, all except 4 succumbed to the disease by the end of follow-up.
We find it noteworthy that high HLA-A, B or C expression by MBC cells was very specific in predicting the 5-year survival. In other words, patients whose MBCs were expressing these molecules in amounts higher than a certain cut-off value were likely to survive at least five years after starting a new line of therapy. We found the specificity of HLA-A at 33.5 40-ΔΔCT to be 87.5% and that of HLA-B/C at 31.9 40-ΔΔCT even reached 100%. It seems that HLA-B/C has a higher AUC value (0.55), while the sensitivities at these cut-off values were identical for both HLA classes (21.1%). However, no clinically useful cut-off points could be determined that would optimize sensitivity, which would help identify patients likely to die of their BC before the 5-year mark, since the highest sensitivity that could be achieved was 73.7% when using a cut-off of 30.5 40-ΔΔCT for HLA-A expression.
The highest diagnostic efficacy, i.e., the compromise between sensitivity and specificity, is represented by the ROC Youden J point cut-off values of 33.5 40-ΔΔCT for HLA-A and 31.9 40-ΔΔCT for HLA-B/C. To the best of our knowledge, no other study has thus far dealt with the topic of HLA class I molecule cut-off points in the context of predicting survival in MBC patients.
We could make no inference in regard to positive and negative predictive values because the five-year survival has never yet been statistically evaluated in a group of patients comparable to those enrolled in our study–patients with MBC with high HLA-A or HLA-B/C expression. This particular aspect warrants further studies.
We acknowledge that our study has several limitations. First and foremost, the number of patients enrolled is relatively low thus making our conclusions less reliable - 34 patients were enrolled in the study overall and only 7 patients survived to five years or longer. However, the methods applied in this study are resource-consuming, including mRNA-based determination of intrinsic tumor subtypes and HLA molecules of several classes. The applied methods require highly specialized skillsets and are associated with significant costs. Thus, this was a proof-of-concept study, and we plan to analyze a larger patient cohort in the future since determining HLA-A, B or C expression might be able to predict which patients would benefit most from expensive new-line MBC treatments. Second, not all patients with high levels of HLA-A, B or C expression by their MBC cells had favorable outcomes, so future research should be undertaken to define other predictors of such events.
Conclusion
Our proof-of-concept study supports the conclusion that high expression of certain class I HLA molecule subtypes by MBC cells can identify patients with a favorable prognosis. Our data suggest cut-off values for HLA-A and HLA-B/C expression with high specificity as a potential prognosticator of a 5-year overall survival, paired with a yet too low sensitivity based on our pilot cohort. Therefore, additional prospective follow-up studies are required.
Footnotes
Authors’ Contributions
Conception and design of the study: S.S., M.W., T.M.D.; Data collection: R.W., A.S.; Data analysis & interpretation: U.K., S.S., M.S.; Statistical analysis: U.K., S.S.; Manuscript preparation phase 1 - drafting the article: S.S., U.K., T.M.D.; Manuscript preparation phase 2- revising it critically for important intellectual content: M.W., M.S., R.W., A.S.; Final approval of the version to be submitted: all Authors.
Conflicts of Interest
R. W. is an employee of Stratifyer Molecular Pathology GmbH. All other Authors have nothing to disclose in relation to this study.
- Received January 18, 2023.
- Revision received February 1, 2023.
- Accepted February 2, 2023.
- Copyright © 2023 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- Correlation of HLA-A and HLA-B/C Expression With PD-1 and PD-L1 Expression in Patients With Metastatic Breast Cancer as a Potential Prognosticator of Favorable Survival
- Correlation of HLA-A and HLA-B/C Expression With ESR1 Expression in Patients With Metastatic Breast Cancer as a Potential Prognosticator of Favorable Distant Disease-free Survival