University of Groningen
Tumor immunology in ovarian cancer
Merkus-Brunekreeft, Kim
DOI:
10.33612/diss.147014180
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Merkus-Brunekreeft, K. (2020). Tumor immunology in ovarian cancer. University of Groningen.
https://doi.org/10.33612/diss.147014180
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
7
177 176
Summary
7
SUMMARY
Treatment of ovarian cancer has been improved and standardized over the last years. Major developments have been the focus on achieving complete removal of macroscopic tumor remains and; the implementation of neoadjuvant chemotherapy. More recently, maintenance therapy with poly (ADP-ribose) polymerase inhibitors (PARP-inhibitors) has come forward as a novel approach for reducing relapse rate. Nevertheless, overall survival has barely improved and ovarian cancer remains the deadliest gynecological malignancy worldwide. The majority of ovarian cancers is diagnosed at a late stage of disease and chemotherapeutic therapies are associated with high-toxicity. Moreover, recurring ovarian cancer is often chemotherapy-resistant. Altogether, these features underline the importance to develop an alternative approach. One promising approach that has recently gained considerable attention is immunotherapy. In this thesis we investigated several immune cell compartments in ovarian cancer patients including the tumor itself, peripheral blood and draining lymph nodes.
In chapter 2, 3 and 4, we show that ovarian cancer tumors are characterized by activated, but exhausted immune cell populations, while tumor-draining lymph nodes are characterized by a quiescent immune cell signature. Our data suggest a lack of adequate priming of immune cells in the lymph nodes, possibly because of a lack of tumor-specific antigen presentation, resulting in a quiescent immune cell population unable to induce any cytotoxicity. In chapter 5 we thus propose a strategy to augment T cell activity via the delivery of CD40L to cancer cells to activate antigen-presenting cells in tumor-infiltrated tumor-draining lymph nodes. We show that binding of these protein-constructs led to the in vitro activation of dendritic cells and T cell proliferation, but is not yet tested in a preclinical in vivo model. As there is currently no adequate preclinical in vivo model that represents the complexity of the human immune system and the human ovarian tumor to validate immune system molding strategies like previously described, we aimed to develop an immune-humanized patient-derived xenograft model harboring both components. In chapter 6 we show that it is possible to establish an ovarian cancer patient derived xenograft model using ectopic tumor tissue implantation. However, the incorporation of human immune system components remained challenging.
In addition to the limited activation of tumor-draining lymph nodes, we saw a significant decrease of the myeloid cell population in the peripheral blood of
high-grade serous ovarian cancer (HGSOC) patients during treatment with neoadjuvant chemotherapy, with a partial recovery upon completion of the chemotherapy. Therefore, chemotherapy treatment might potentiate immunotherapy through the depletion of myeloid cells, which are mostly known for their immunosuppressive properties. By contrast, we observed in chapter 3 that tumor-infiltrating immune cells appeared only to be of prognostic benefit in patients receiving primary debulking surgery and not in those who received neoadjuvant chemotherapy and thereafter chemotherapy. We show in chapter 4 that this might be explained by varying expression of major histocompatibility class I (MHC-I) in the two different patient groups. HGSOC tumors exposed to neoadjuvant chemotherapy are characterized by infrequent expression of MHC-I, in contrast to HGSOC tumors that are chemotherapy-naive expressing high levels of MHC-I. A prognostic benefit of tumor-infiltrating lymphocytes was observed in the chemotherapy-naïve tumor group, indicating the necessity of antigen presentation via MHC-I for activation of the immune system. To establish that these patients did indeed show tumor-reactive T cell infiltration that was inhibited by lack of antigen presentation, we established a platform of in vitro (e.g. cell cultures, polymerase chain reaction) and in silico (e.g. bioinformatics) identification of tumor-specific neoantigens. In chapter 6 we used this platform to identify neoantigen specific T cell responses for ovarian cancer patients.
DISCUSSION AND CONCLUSION
This thesis consists of fundamental studies on the immune cell populations in various immune cell compartments of ovarian cancer patients: the tumor, the peripheral blood, and draining lymph nodes. Moreover, the development of an in vivo model for the validation of novel immunotherapeutic strategies for ovarian cancer is discussed. In this section the research discussed in this thesis is related to current knowledge and developments within this area.
Tumor-draining lymph nodes as site of immune silencing in ovarian cancer.
7
prior to chemotherapy. Not just lymphocytes itself, also their differentiation status and location are crucial since not all immune cell subsets contribute to prognosis in the same manner. In general, the immune cell populations present within the tumor have been most extensively studied. We expanded the existing data by including both markers for immune cell localization (CD103) and by performing a side-by-side evaluation of the immune contexture of tumor in both primary and neoadjuvant treated tumors, using both flow cytometry of primary fresh samples and retrospective tissue archived cohorts. While these data provided extensive novel insight, one caveat of the retrospective analysis is the use of tissue microarrays (TMAs). A major part of the data presented in chapter 2 and 3 is gathered using consecutive TMA-slides. TMAs contain small representative tissue samples of many different patients on one histological slide and aid to identify new diagnostic and prognostic markers and/ or targets in human cancers.1 TMAs are the solution to the otherwise costly,
time-consuming and laborious histological techniques. Although there only seems to be ample advantages of TMAs, one major concern may be the representative value of the small tissue samples to the whole tumor. Camp et al. and Parker et al. used TMAs to investigate respectively the proliferation status and estrogen receptor status in breast cancer and found 95-96% concordance compared to whole slide staining.2,3 Camp
et al. revealed that analysis of 2 cores per sample resulted in 95% concordance.2
Nocito et al. proved the same for the proliferation status and histological grade of bladder cancer.4 Another point of criticism could be the presence of only one core
per sample per microarray, in our study 3 cores per patient were present on a slide. A sample was only included in further analyses when two or more cores could be subjected to analysis, meaning a minimum of 20% of the core had to consist of tumor cells. Of course, still any core and/or sample could be completely negative for tumor cells or absent on any given microarray. In our approach, however, we analyzed many samples of an individual patient simultaneously to eliminate this bias. Finally, our side-by-side analysis of both fresh tissue and full slides from the same patients as well as comparison of full slides by immunofluorescence and TMA cores of the same patients provide a highly correlated measurement, at least for T cells. While not exhaustively examined in detail here, considerable immune cell heterogeneity for TMAs has so far not been reported.
While the focus of most tumor immunology studies in ovarian cancer patients has been on the level of the tumor, most immune cells circulate between blood, lymph nodes and tissue. Therefore, we also investigated the immune status of these compartments to broaden our understanding of the immune system in ovarian cancer patients.
Most research on lymph nodes within cancer patients has been done on sentinel lymph node resection and biopsies, in particular in breast cancer.5,6 Only a few studies
have been done on the immune status of these lymph nodes.7,8 Here, like in cervical
cancer, a quiescent immune cell population dominated the immune cells in the draining lymph nodes of ovarian cancer patients9, even when there was adequate
MHC-I expression within the tumor. It is tempting to speculate that the inadequate presentation of antigens within the draining lymph nodes explains the relatively low responses to immune checkpoint inhibition of ovarian cancer patients. Indeed, mouse models have demonstrated a significant upregulation of PD-L1 in the draining lymph nodes of tumors when compared to the non-draining lymph nodes, a phenomenon suggesting active tumor-induced immune suppression of the initial priming.10 More
importantly, when mice responsive to immune checkpoint inhibitors were subjected to a surgical resection of their draining lymph node, the therapeutic effect was completely abolished. These findings could be confirmed using a chemical inhibitor of T cell circulation, in line with our hypothesis that parallel evaluation of the various immune compartments is an essential feature of immune profiling. Future studies should thus focus on ensuring adequate tumor-specific antigen presentation to induce a proper immune response in the draining lymph nodes of ovarian cancer.
Of note, a potential caveat of the current study was that the draining lymph nodes used in chapter 2 were visually identified by the surgeon (e.g. enlarged lymph nodes in proximity of the tumor), however their tumor draining status was not confirmed before resection. To reassure their draining status, prospective clinical studies should be performed in which the draining status of a lymph node is confirmed using a dye or tracer prior to resection.11 Nevertheless, we observed tumor-infiltration in a number
of nodes consistent with their draining status.
As we identified a quiescent immune microenvironment in the draining lymph node which might explain the poor immune response to the tumor in ovarian cancer, future combination immunotherapy strategies should focus on creating a favorable immune lymph node microenvironment, such as vaccination or dendritic cell activating regimes. In chapter 5 we discuss one such approach with an in vitro validated strategy which enhances dendritic cell (DC) maturation locally by targeted delivery of the tumor necrosis factor family member CD40 ligand. CD40 is a crucial costimulatory molecule on DCs and a target for several agonistic monoclonal antibodies currently under active clinical investigation. Unfortunately, most of these antibodies suffer from an unfavorable toxicity profile, most notably severe liver toxicity. As a result, more targeted approaches such as those conferred by antibody-based fusions
181 180
Summary
7
(immunocytokines) would be of benefit. Several immunocytokines are currently under clinical development, mainly aimed at PD-L1 and/or CD25 (the IL-2 receptor). While traditionally more difficult to produce for clinical application, these molecules offer considerable advances in terms of local activity and toxicity profile. As discussed in chapter 5, CD40L activity could be almost completely made conditional on the presence of the target antigen molecule (EpCAM or CD20). As these molecules are site-specific and active at lower concentration, for clinical translation, could therefore envision bypassing the in vitro production and pursue an in vivo production step, such as that recently proposed for an mRNA platform.
Taken together, advances at the priming stage of the immune response may be key to overcoming immune dysfunction in ovarian cancer.
Standard-of-care treatment and immunotherapy in ovarian cancer
In addition to the above, our data have identified intriguing connections between the standard-of-care treatment and immune contexture of ovarian cancer. Currently the standard of care for ovarian cancer consists of chemotherapy combined with debulking surgery. Primary debulking surgery prior to six three-week cycles of platinum-based chemotherapy is the standard. However, interval debulking is increasingly chosen over primary debulking surgery because it is associated with less surgery-induced morbidity and improved surgical outcomes. Additionally, different ways of administration of chemotherapy gained interest. Intravenous chemotherapy is still the first choice, but intraperitoneal chemotherapy gains more interest, in particular hyperthermic intraperitoneal chemotherapy (HIPEC). HIPEC leads to a longer disease-free period and overall survival in patients that have undergone complete or optimal debulking surgery and no higher level of toxicities.12 Recurrent ovarian cancers are
in the end often platinum-resistant, these tumors can be treated with various other chemotherapies, e.g. topotecan, doxorubicin, gemcitabine. However results are marginal, only 12-15% percent of all patients will respond. Only one in three patients will have stable disease. The majority will still show progression of disease within 3-6 months.13,14
In chapter 2 we show that neoadjuvant chemotherapy does in fact decrease the number of MDSC present in the peripheral blood and does not affect the T cell populations present, this window of opportunity could be used to treat ovarian cancer with T cell mediated immunotherapy. A randomized controlled trial is currently planned which aims to evaluate the efficacy of Pembrolizumab in combination with standard of care treatment in patients with advanced ovarian cancer. Assuming that
its administration will improve immune response leading to optimal surgery results (NCT03275506; estimated primary completion date: September 2021). However, a note of caution based on our data is warranted. As the beneficial effects of immune checkpoint inhibitors appear to be dependent, at least in part, on the involvement of the tumor-draining lymph node, the combination of pembrolizumab with chemotherapy may prove ineffective without proper up-front priming of the response. An alternative may therefore be the inclusion of a tumor antigen-specific vaccination into first-line chemotherapy treatment, followed by immune checkpoint inhibitor (maintenance) treatment. An ongoing study aims to address this approach by evaluating the combination of an mRNA vaccination during first-line chemotherapy treatment in ovarian cancer (2017-004585-10). The outcome of the trial may help development of follow-up randomized trials including immune checkpoint inhibitors.
Altogether, it is anticipated that combined chemo-immunotherapy will take center stage in ovarian cancer treatment for years to come.
Alternatives for the ovarian cancer patient-derived xenograft model
The majority of research in ovarian cancer is done in the syngeneic ID8 model and derivates of this model.15-17 But this model does not represent the heterogeneity of
disease. Other alternatives to murine models would be the egg-laying hen or the jaguar.22
Next to humans they are the only animals known for the spontaneous development of ovarian cancer, however using them for research purposes is far from feasible. These are laborious and time-consuming models, being an endangered species, and there are no reagents available for these species.18,19-21 With these obstacles in mind
and increasing demands for animal-free drug development alternatives, Mucaki et al. designed an in silico method that accurately predicted 60.2% of disease recurrence and 61% of remission in ovarian cancer based on gene signatures and their response to chemotherapy.23 Although these developments are constantly improved they are
not that developed they can replace in vivo models. Another point of critique is the low predictable value and the irreproducibility of preclinical research, which is reported to be between 51-89%.23 Illustrating the need for improved design of preclinical in vivo
models, and at the same time the importance of animal welfare.24-26 In chapter 6, we
analyzed the potential of a patient-derived xenograft (PDX) model for modeling anti-tumor immunity. PDX models capture the heterogeneity of ovarian cancer and can in principle be combined with autologous immune cells in an adoptive cell transfer setting, or using HLA-matched cord blood humanized mice. Indeed, a recent study by Want et al. performed neoantigen profiling of ovarian cancer PDX and found a remarkable
7
correlation between primary tumor material and PDX material in terms of mutations.27
These findings are in line with data from the group of de Jong establishing a conserved profile of mutations, copy number alterations and methylation profile of ovarian cancer PDX vs. primary tumors.28 Accordingly, Want et al. were able to demonstrate specific
T cell responses against neoantigens present within the autologous PDX. Adoptive cell transfer (ACT) of neoantigen-specific clones resulted in potent tumor growth inhibition, suggesting the feasibility of an ACT approach for ovarian cancer treatment. Importantly, while the frequency of neoantigen specific T cells in peripheral blood was observed in only a subset of patients, Bobisse et al. demonstrate that parallel evaluation of both PBMC and TIL allowed neoantigen-specific T cell isolation in almost all patients tested, despite a low number of somatic mutations in these tumors opening the way to personalized immunotherapy for treatment of ovarian cancer.29
The optimal treatment for every patient
With the explosion of all novel immunotherapeutic options and their increasing specificity the necessity arises to find the optimal treatment for every individual patient with ovarian cancer: To obtain optimal treatment results but also to reduce costs and become cost-effective. This can be done by developing methods to select the right treatment for each patient (decision-making). In vitro models often fall short and most preclinical in vitro models are too time-consuming to serve clinical purposes therefore alternative approaches are sought. For example, the CIVO tumor microinjection platform, a handheld device that can hold up to eight needles capable of simultaneously delivering micro-dosages of potential therapeutics to subcutaneous tumors.30 A different device, however with the same rationale, was demonstrated
by Jonas et al., an implantable device holding up to 16 reservoirs with therapeutics enabling high-throughput of in vivo testing.31 These devices are only suitable for
subcutaneous tumors and in need of physical present tumor. Therefore, in silico
prediction models gain interest to circumvent the need of tumors. Sun et al. designed a 16-gene signature platform (IndividCRS), based on cell lines and ovarian cancer tumors from 25 independent datasets, to not only monitor response to chemotherapy but also predict the tumor sensitivity to chemotherapy.32 But like the previous mentioned
method of Mucaki et al. both need to be improved before any clinical implications.33,34
However, we approached it from another way: designing a platform enabling the development of unique therapeutics for every individual patient, this approach is also known as: personalized medicine. We designed a platform to identify neoantigens specific to the tumor of individual patients and thereafter select and expand
neoantigen-specific T cells. In chapter 6 we discussed our strategy to select a patient-specific immunotherapy, using various sequential techniques, including cell cultures, polymerase chain reaction, flow cytometry and bioinformatics. Unfortunately, we were not able to establish an immune competent ovarian cancer patient-derived xenograft. Nevertheless, as mentioned, Bobisse et al. managed to identify high avidity neoepitope specific CD8+ T cells in immunotherapy naïve ovarian cancer patients,
showing the possibility to apply this also to tumors with a low mutational load, like ovarian cancer.29
Conclusion
The data presented in this thesis show that the immune system plays without a doubt a role in ovarian cancer, but it is influenced by the presence of the tumor and affected by the administered chemotherapy. In order to return its antitumor properties, an immune-activating therapeutic that enables the immune system to recognize and destroy the ovarian cancer cells is highly desired. Novel immunotherapies are widely investigated in vitro but to be validated in an immune-competent in vivo model for ovarian cancer and implemented within the current standard of care. However, a solid immune-competent in vivo ovarian cancer model is yet to be found. All efforts with the final aim to improve the disease free and overall survival of ovarian cancer patients.
185 184
Summary
7
REFERENCES
1. Jawhar NM. Tissue Microarray: A rapidly evolving diagnostic and research tool. Ann Saudi Med 2009 Mar-Apr;29(2):123-127.
2. Camp RL, Charette LA, Rimm DL. Validation of tissue microarray technology in breast carcinoma. Lab Invest 2000 Dec;80(12):1943-1949.
3. Parker JS, Mullins M, Cheang MC, Leung S, Voduc D, Vickery T, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol 2009 Mar 10;27(8):1160-1167.
4. Nocito A, Bubendorf L, Tinner EM, Suess K, Wagner U, Forster T, et al. Microarrays of bladder cancer tissue are highly representative
5. Bouquet de Joliniere J, Major A, Khomsi F, Ben Ali N, Guillou L, Feki A. The Sentinel Lymph Node in Breast Cancer: Problems Posed by Examination During Surgery. A Review of Current Literature and Management. Front Surg 2018 Nov 14;5:56.
6. Zahoor S, Haji A, Battoo A, Qurieshi M, Mir W, Shah M. Sentinel Lymph Node Biopsy in Breast Cancer: A Clinical Review and Update. J Breast Cancer 2017 Sep;20(3):217-227.
7. Takada K, Kashiwagi S, Goto W, Asano Y, Takahashi K, Fujita H, et al. Possibility of avoiding axillary lymph node dissection by immune microenvironment monitoring in preoperative chemotherapy for breast cancer. J Transl Med 2018 Nov 19;16(1):318-018-1692-3.
8. Denkert C, von Minckwitz G, Brase JC, Sinn BV, Gade S, Kronenwett R, et al. Tumor-infiltrating lymphocytes and response to neoadjuvant chemotherapy with or without carboplatin in human epidermal growth factor receptor 2-positive and triple-negative primary breast cancers. J Clin Oncol 2015 Mar 20;33(9):983-991.
9. Heeren AM, Koster BD, Samuels S, Ferns DM, Chondronasiou D, Kenter GG, et al. High and interrelated rates of PD-L1+CD14+ antigen-presenting cells and regulatory T cells mark the
microenvironment of metastatic lymph nodes from patients with cervical cancer. Cancer Immunol Res 2015 Jan;3(1):48-58.
10. Fransen MF, Schoonderwoerd M, Knopf P, Camps MG, Hawinkels LJ, Kneilling M, et al. Tumor-draining lymph nodes are pivotal in PD-1/PD-L1 checkpoint therapy. JCI Insight 2018 Dec 6;3(23):10.1172/jci.insight.124507.
11. Kleppe M, Brans B, Van Gorp T, Slangen BF, Kruse AJ, Pooters IN, et al. The detection of sentinel nodes in ovarian cancer: a feasibility study. J Nucl Med 2014 Nov;55(11):1799-1804.
12. of proliferation index and histological grade. J Pathol 2001 Jul;194(3):349-357.
13. van Driel WJ, Koole SN, Sikorska K, Schagen van Leeuwen JH, Schreuder HWR, Hermans RHM, et al. Hyperthermic Intraperitoneal Chemotherapy in Ovarian Cancer. N Engl J Med 2018 Jan 18;378(3):230-240.
14. Soyama H, Takano M, Miyamoto M, Yoshikawa T, Aoyama T, Goto T, et al. Factors favouring long-term survival following recurrence in ovarian cancer. Mol Clin Oncol 2017 Jul;7(1):42-46.
15. Pignata S, C Cecere S, Du Bois A, Harter P, Heitz F. Treatment of recurrent ovarian cancer. Ann Oncol 2017 Nov 1;28(suppl_8):viii51-viii56.
16. Fong MY, Kakar SS. Ovarian cancer mouse models: a summary of current models and their limitations. J Ovarian Res 2009 Sep 28;2(1):12-2215-2-12.
17. Roby KF, Taylor CC, Sweetwood JP, Cheng Y, Pace JL, Tawfik O, et al. Development of a syngeneic mouse model for events related to ovarian cancer. Carcinogenesis 2000 Apr;21(4):585-591.
18. Quinn BA, Xiao F, Bickel L, Martin L, Hua X, Klein-Szanto A, et al. Development of a syngeneic mouse model of epithelial ovarian cancer. J Ovarian Res 2010 Oct 19;3:24-2215-3-24.
19. McCloskey CW, Rodriguez GM, Galpin KJC, Vanderhyden BC. Ovarian Cancer Immunotherapy: Preclinical Models and Emerging Therapeutics. Cancers (Basel) 2018 Jul 26;10(8):10.3390/ cancers10080244.
20. Johnson PA, Giles JR. The hen as a model of ovarian cancer. Nat Rev Cancer 2013 Jun;13(6):432-436.
21. Corner S, Parys M, Moresco A, Yuzbasiyan V, Agnew D. Ovarian adenocarcinoma in jaguars (panthera once): Naturally-occuring model of human inherited ovarian cancer? Doi: 10.1158/1557-3265.OVCASYMP16-TMEM-020. Published June 2017.
22. Hawkridge AM. The chicken model of spontaneous ovarian cancer. Proteomics Clin Appl 2014 Oct;8(9-10):689-699.
23. Mucaki EJ, Zhao JZL, Lizotte DJ, Rogan PK. Predicting responses to platin chemotherapy agents with biochemically-inspired machine learning. Signal Transduct Target Ther 2019 Jan 11;4:1-018-0034-5. eCollection 2019.
24. Singh VP, Pratap K, Sinha J, Desiraju K, Bahal D, Kukreti R. Critical evaluation of challenges and future use of animals in experimentation for biomedical research. Int J Immunopathol Pharmacol 2016 Dec;29(4):551-561.
25. Freedman LP, Cockburn IM, Simcoe TS. The Economics of Reproducibility in Preclinical Research. PLoS Biol 2015 Jun 9;13(6):e1002165.
26. Halsey LG, Curran-Everett D, Vowler SL, Drummond GB. The fickle P value generates irreproducible results. Nat Methods 2015 Mar;12(3):179-185.
27. Begley CG, Ioannidis JP. Reproducibility in science: improving the standard for basic and preclinical research. Circ Res 2015 Jan 2;116(1):116-126.
28. Want MY, Konstorum A, Huang RY, Jain V, Matsueda S, Tsuji T, et al. Neoantigens retention in patient derived xenograft models mediates autologous T cells activation in ovarian cancer. Oncoimmunology 2019 Mar 30;8(6):e1586042.
29. Alkema NG, Tomar T, Duiker EW, Jan Meersma G, Klip H, van der Zee AG, et al. Biobanking of patient and patient-derived xenograft ovarian tumour tissue: efficient preservation with low and high fetal calf serum based methods. Sci Rep 2015 Oct 6;5:14495.
30. Bobisse S, Genolet R, Roberti A, Tanyi JL, Racle J, Stevenson BJ, et al. Sensitive and frequent identification of high avidity neo-epitope specific CD8 () T cells in immunotherapy-naive ovarian cancer. Nat Commun 2018 Mar 15;9(1):1092-018-03301-0.
31. Klinghoffer RA, Bahrami SB, Hatton BA, Frazier JP, Moreno-Gonzalez A, Strand AD, et al. A technology platform to assess multiple cancer agents simultaneously within a patient’s tumor. Sci Transl Med 2015 Apr 22;7(284):284ra58.
32. Jonas O, Landry HM, Fuller JE, Santini JT,Jr, Baselga J, Tepper RI, et al. An implantable microdevice to perform high-throughput in vivo drug sensitivity testing in tumors. Sci Transl Med 2015 Apr 22;7(284):284ra57.
33. Sun J, Bao S, Xu D, Zhang Y, Su J, Liu J, et al. Large-scale integrated analysis of ovarian cancer tumors and cell lines identifies an individualized gene expression signature for predicting response to platinum-based chemotherapy. Cell Death Dis 2019 Sep 10;10(9):661-019-1874-9.
34. Mucaki EJ, Zhao JZL, Lizotte DJ, Rogan PK. Predicting responses to platin chemotherapy agents with biochemically-inspired machine learning. Signal Transduct Target Ther 2019 Jan 11;4:1-018-0034-5. eCollection 2019.
