The CD47/TSP-1 axis: a promising avenue for ovarian cancer treatment and biomarker research

Abstract

Background: Ovarian cancer (OC) remains one of the most challenging and deadly malignancies facing women today. While PARP inhibitors (PARPis) have transformed the treatment landscape for women with advanced OC, many patients will relapse and the PARPi-resistant setting is an area of unmet medical need. Traditional immunotherapies targeting PD-1/PD-L1 have failed to show any benefit in OC. The CD47/TSP-1 axis may be relevant in OC. We aimed to describe changes in CD47 expression with platinum therapy and their relationship with immune features and prognosis.

Methods: Tumor and blood samples collected from OC patients in the CHIVA trial were assessed for CD47 and TSP-1 before and after neoadjuvant chemotherapy (NACT) and multiplex analysis was used to investigate immune markers. Considering the therapeutic relevance of targeting the CD47/TSP-1 axis, we used the CD47-derived TAX2 peptide to selectively antagonize it in a preclinical model of aggressive ovarian carcinoma.

Results: Significant reductions in CD47 expression were observed post NACT. Tumor patients having the highest CD47 expression profile at baseline showed the greatest CD4⁺ and CD8⁺ T-cell influx post NACT and displayed a better prognosis. In addition, TSP-1 plasma levels decreased significantly under NACT, and high TSP-1 was associated with a worse prognosis. We demonstrated that TAX2 exhibited a selective and favorable biodistribution profile in mice, localizing at the tumor sites. Using a relevant peritoneal carcinomatosis model displaying PARPi resistance, we demonstrated that post-olaparib (post-PARPi) administration of TAX2 significantly reduced tumor burden and prolonged survival. Remarkably, TAX2 used sequentially was also able to increase animal survival even under treatment conditions allowing olaparib efficacy.

Conclusions: Our study thus (1) proposes a CD47-based stratification of patients who may be most likely to benefit from postoperative immunotherapy, and (2) suggests that TAX2 is a potential alternative therapy for patients relapsing on PARP inhibitors.

Keywords: CD47; Neoadjuvant chemotherapy; Ovarian cancer; PARP; TSP-1.

Fig. 1

Schematic illustration of two major clinical challenges in ovarian carcinoma. The lack of therapeutic response to immune checkpoint inhibitors (ICI, top panel). There is no treatment for patients who relapse after treatment with PARPi (bottom panel)

Fig. 2

Immunohistochemical assessment of CD47 expression in ovarian carcinomas at diagnosis and after neoadjuvant chemotherapy. CD47 expression was assessed by IHC in paired samples at diagnosis (baseline, green point) and after neoadjuvant chemotherapy (NACT) (orange triangle) (n = 65). The quantification was conducted using the H-score system, combining staining intensity (0, + 1, + 2, + 3) with the percentage of positive cells (0 to 100) for a score of 0 to 300. a Lines connect samples from the same patient and indicate an increase (red) or decrease (green) in CD47 expression (left panel). Histograms showing quantification of CD47 expression (right panel). Statistical significance is indicated as **p < 0.01 (two-tailed Mann–Whitney test). b Representative images displaying representative CD47 staining patterns in three tumors from OC patients before and after NACT

Fig. 3

(previous page). Correlation of CD47 with immune features after NACT. Tumor samples were collected from 101 patients from the CHIVA trial. a At baseline, we defined a CD47^highest (fourth quartile, score above 240) and a CD47^lowest population (first quartile, score under 150). Tumor samples were collected from 101 patients from the CHIVA trial. CD4⁺, CD8⁺, CD68⁺, CD163⁺, and FoxP3⁺ cells were assessed by immunofluorescence and scored as the number of positive cells. A mean score was calculated from three TMA cores from each sample. b CD8 (green point) and CD4 (blue square) staining were quantified at baseline (upper panels) or post NACT (lower panels) in the CD47^lowest (left panels) and CD47^highest (right panels) populations. The Pearson correlation coefficient (r) is indicated on each panel. The p-value was determined using the Wald test. c, Histogram representing CD4⁺ (blue), CD8⁺ (green), and FOXP3⁺ (pink) signal intensities at baseline (circle) and after NACT (triangle) in the CD47^lowest and CD47.^highest populations. Statistical significance is indicated as ns (not significant) or *p < 0.05 (one-tailed Mann–Whitney test)

Fig. 4

Circulating TSP-1 is decreased after NACT. a Schematic illustration depicting the functional interaction between the CD47 receptor and the TSP-1 glycoprotein (left panel). The strategy employed to measure circulating TSP-1 levels in OC patients is depicted in the right panel. b Plasma levels of TSP-1 were determined using paired samples (n = 59) from the CHIVA cohort at baseline (green point) and post NACT (blue triangles). Connective lines (red for increased, green for decreased) link samples from the same patient to illustrate changes after treatment. Statistical significance is indicated by ****p < 0.0001 (two-tailed Mann–Whitney test). c Correlation plot displaying plasma TSP-1 levels versus tissue CD47 expression. d Kaplan–Meier survival curve representing patient overall survival (OS) based on circulating TSP-1 levels. The median was used as a cutoff to discriminate between low and high TSP-1 expression

Fig. 5

TAX2: binding, pharmacokinetics, and biodistribution studies. a Schematic diagram illustrating the mechanism of action of TAX2, which binds the carboxy-terminal domain of TSP-1, selectively inhibiting TSP-1/CD47 interaction. b Graph displaying the dose–response curve for the binding between TAX2 and labeled rhTSP-1 (10⁻⁸ M) as assessed by microscale thermophoresis. c Evaluation of TAX2 biodistribution in healthy C57BL/6 J mice using FMT after IV administration of a dose range of TAX2-Cy5 (1, 5, 10, or 20 mg/kg, 2–6 mice/group). Fluorescence quantification was performed on collected tissues at T0. The middle part of panel c presents the TSP-1 gene (THBS1) expression in various human tissues obtained from the Human Protein Atlas database, combining datasets from HPA, GTEx, and FANTOM5 as of April 2024. The table (lower part of panel) presents fluorescent quantification on plasma collected at T0. d Assessment of TAX2-Cy5 biodistribution (administered IV at a 10 mg/kg BW dose) in mice bearing ID8 Trp53^−/− Brca2.^−/− ovarian carcinoma cells injected either subcutaneously (SC) or intraperitoneally (IP). Fluorescence quantification using FMT was conducted on live mice after 24 h (bottom panel). Subsequently, subcutaneous tumors (black bar, SC model), peritoneal tumor masses, and metastatic nodules (pink and brown bars, respectively, IP model) were excised post euthanasia and then imaged for TAX2-Cy5 quantification. Necropsy (bottom panel, right image) reveals metastases in the omentum (highlighted by white arrows)

Fig. 6

TAX2 efficacy in a metastatic mouse model of ovarian cancer. a A metastatic ovarian carcinoma model was established in C57BL/6 J female mice by IP injection of 2 × 10⁶ ID8 Trp53^−/− Brca2^−/− cells. Tumor burden was monitored by assessing the change in relative body weight as a result of intense ascites production over 6 weeks post inoculation with tumor cells. Plasma TSP-1 concentration was quantified using an ELISA assay. b C57BL/6 J female mice (n = 9–19 per group) were IP inoculated with ID8 Trp53^−/− Brca2^−/− cells and treated the following day with either olaparib (50 mg/kg, per os, daily for 2 weeks) or TAX2 (30 mg/kg, IV, 3 times weekly for 6 weeks). Kaplan–Meier overall survival analysis illustrates the outcome among control mice (black), olaparib (orange), and TAX2 treated mice (blue). Survival curves were compared using the Gehan–Breslow–Wilcoxon test as previously described [37]. The table presents median survival for each treatment group. c C57BL/6 J female mice were IP injected with ID8 Trp53^−/− Brca2.^−/− and treated with vehicle (n = 9), olaparib alone (50 mg/kg, per os, daily for 2 weeks, n = 20) or sequentially with olaparib (50 mg/kg, per os, daily for 2 weeks) followed by TAX2 (30 to 100 mg/kg, IV, 3 times weekly for 4 weeks) (n = 8 per TAX2 dosing). The left panel indicates the tumor response, with the dotted black circle representing mice unresponsive to olaparib treatment. The right panel presents Kaplan–Meier overall survival analysis comparing control mice (black, n = 9), olaparib-resistant mice (orange, n = 14), and mice treated sequentially with olaparib then TAX2 (30 mg/kg, green, n = 19), using the Gehan–Breslow–Wilcoxon test. *p < 0.05, ***p < 0.001

Fig. 7

Efficacy of a sequential combination of olaparib with TAX2 in a peritoneal carcinomatosis model. ID8 Trp53^−/− Brca2.^−/− carcinoma cells were IP injected into mice as described in Fig. 6 to induce peritoneal carcinomatosis. At day 7 post inoculation, animals were randomly assigned (n = 7–11 per group) and treated with olaparib for 5 weeks (per os, daily at 50 mg/kg) or an equivalent volume of olaparib vehicle. Subsequently, mice were sequentially treated with TAX2 until the end of the protocol (IV, three times a week at 30 mg/kg). Kaplan–Meier overall survival analysis comparing control mice (black), olaparib-vehicle mice (gray), olaparib-treated mice (orange), and mice treated sequentially with olaparib then TAX2 (green) is presented. Statistical analysis was performed using the Gehan–Breslow–Wilcoxon test, denoted as *p < 0.05 and ***p < 0.001