Ovarian Cancer Drug Shows Eye and Nerve Risks in Large Real-World Study

A promising antibody-drug conjugate for hard-to-treat ovarian cancer is raising new safety questions, according to one of the largest real-world analyses of the drug published to date. Researchers in Taiwan mined a federated electronic health record network spanning dozens of healthcare organizations to compare mirvetuximab soravtansine against conventional salvage chemotherapy in women with platinum-resistant ovarian cancer. Their findings, published in the Journal of Ovarian Research, paint a nuanced picture: a therapy that reached patients late in the disease course carried a clearly elevated risk of two distinctive side effects, corneal disorders and peripheral neuropathy, even as overall survival remained statistically indistinguishable between the two treatment groups.

Mirvetuximab soravtansine belongs to a class of engineered molecules called antibody-drug conjugates, which are often described as biological guided missiles. The antibody portion of the molecule is designed to bind folate receptor alpha, a protein abundantly displayed on the surface of many epithelial ovarian cancer cells but largely absent from most healthy tissues. Once the antibody docks onto its target, the cancer cell engulfs the construct, and a potent cytotoxic payload, a derivative of the maytansinoid family that disrupts microtubule assembly, is released inside the cell. In principle, this strategy delivers chemotherapy directly to tumor tissue while sparing normal cells, an idea that has transformed treatment in breast cancer and other malignancies. In practice, the approach is never perfectly selective, and the new study quantifies precisely where the collateral damage appears in routine clinical care.

The clinical problem the drug addresses is one of the most stubborn in oncology. Platinum-resistant ovarian cancer, together with its fallopian tube and primary peritoneal counterparts, emerges when the disease relapses within roughly six months of a platinum-based regimen. At that point, options narrow to non-platinum salvage chemotherapies such as pegylated liposomal doxorubicin, paclitaxel, or topotecan, all of which offer modest response rates and meaningful toxicity. Regulatory approval of mirvetuximab soravtansine opened a biomarker-guided alternative, but randomized trials enroll carefully selected patients, and oncologists have lacked a broad view of how the drug is actually being deployed and how it performs in the messy, comorbidity-laden world of community and hospital practice.

To fill that gap, the research team, led by Yu-Wen Sung of China Medical University Hospital in Taichung, turned to the TriNetX Global Collaborative Network, a federated database that aggregates de-identified electronic health records from participating healthcare organizations worldwide. The retrospective observational study included female patients with recurrent ovarian, fallopian tube, or primary peritoneal cancer who initiated either mirvetuximab soravtansine or conventional salvage chemotherapy between November 2022 and March 2026, with documented platinum exposure within the preceding 180 days. From the initial pool of 320 mirvetuximab-treated patients and 1,848 patients who received standard chemotherapy, the investigators used one-to-one nearest-neighbor propensity score matching to construct balanced comparison cohorts of 311 patients each. Matching variables included age, prior exposure to taxanes, bevacizumab, and PARP inhibitors, and the comorbidities hypertension and diabetes, all factors capable of independently influencing toxicity and survival.

The sequencing data reveal a striking pattern in how the drug is entering clinical practice. More than half of the mirvetuximab recipients, 53.8 percent, started the antibody-drug conjugate at the third line of therapy or later, and 7.2 percent did not receive it until after their sixth line of treatment. That distribution confirms that, in the real world, mirvetuximab is functioning largely as a late-line option, deployed after patients have already cycled through platinum agents, taxanes, and other targeted therapies. Such late deployment matters for safety interpretation, because heavily pretreated patients carry cumulative nerve and bone marrow damage that can amplify or confound the toxicities attributed to any new agent.

When the researchers tallied adverse events coded in the medical records, two signals stood out with unmistakable statistical force. Corneal disorders, captured under the diagnostic code H18, occurred in 17 of 303 evaluable mirvetuximab patients, or 5.6 percent, while no cases appeared among matched chemotherapy controls, a difference so pronounced that the p value fell below 0.0001. New-onset peripheral neuropathy, coded as G62, was diagnosed in 28.8 percent of the mirvetuximab cohort compared with 8.9 percent of controls, corresponding to an odds ratio of 4.13 with a 95 percent confidence interval of 2.27 to 7.50. Both findings align with the drug’s known biology: folate receptor alpha is expressed at low levels in some ocular surface tissues, and the maytansinoid payload class is intrinsically neurotoxic, consistent with the neuropathy seen with other tubulin-targeting agents.

Perhaps the most clinically actionable insight came from the study’s 90-day landmark analysis, which dissected the timing of toxicity onset rather than merely counting events. Corneal toxicity clustered early, with an odds ratio of 5.94 comparing early versus late diagnosis periods and a p value of 0.001, supporting a strategy of intensified ophthalmologic surveillance within the first months of treatment. Peripheral neuropathy followed a different trajectory, accumulating steadily over time as its incidence continued to rise through day 365; the early-versus-late odds ratio of 1.78, with a p value of 0.130, did not reach conventional significance, indicating no clear temporal concentration. In other words, an oncologist can expect eye problems to announce themselves soon after treatment begins, while nerve damage behaves like a creeping hazard that demands vigilance for as long as the patient remains on therapy.

Not every safety signal separated the groups. Thrombocytopenia, a depletion of platelets that raises bleeding risk, occurred at nearly identical rates in both cohorts, 18.5 percent versus 17.9 percent with a p value of 0.87, suggesting this common chemotherapy toxicity cannot be attributed to the antibody-drug conjugate. Pneumonitis, an inflammatory lung condition watched closely with many targeted agents, could not be meaningfully evaluated in the dataset because of insufficient case numbers. The survival analysis proved equally sobering: median overall survival was 552 days for mirvetuximab recipients versus 532 days for chemotherapy controls, a hazard ratio of 0.967 with a confidence interval spanning unity and a p value of 0.7889. The authors are careful to note a critical caveat: individual folate receptor alpha expression status was unavailable in the database, meaning the cohort almost certainly included patients whose tumors lacked the drug’s intended target, diluting any true benefit.

That caveat defines the frontier for future work. Randomized trials restricted to biomarker-positive patients have shown antitumor activity for mirvetuximab soravtansine, and this real-world study was never designed to overturn those efficacy findings; rather, it characterizes how the drug is used and how its harms manifest in an unselected population. The authors conclude that the distinct temporal fingerprints of corneal and neurologic toxicity should translate directly into clinical practice: early and repeated ophthalmologic examinations for every patient starting the drug, paired with ongoing neurological monitoring throughout the treatment course. They also call for comparative effectiveness research in cohorts with documented folate receptor alpha expression, where the survival question can finally be answered properly. Until then, the study stands as a reminder that the true personality of a new cancer drug emerges only after it leaves the controlled environment of the clinical trial and meets the full diversity of the patients it was built to help.

Subject of Research: Real-world safety and treatment patterns of mirvetuximab soravtansine in platinum-resistant ovarian cancer

Article Title: Real-world treatment patterns and safety of mirvetuximab soravtansine in patients with platinum-resistant ovarian cancer: a multi-center analysis using the TriNetX Global Collaborative Network

Article References: Sung, Y.-W., Chang, C. Y.-Y., Chang, W.-C., Hung, Y.-C., Lin, W.-C., Yeh, L.-S., & Tsai, K.-D. (2026). Real-world treatment patterns and safety of mirvetuximab soravtansine in patients with platinum-resistant ovarian cancer: a multi-center analysis using the TriNetX Global Collaborative Network. Journal of Ovarian Research. https://doi.org/10.1186/s13048-026-02269-6

Image Credits: AI Generated

DOI: 10.1186/s13048-026-02269-6

Keywords: mirvetuximab soravtansine, platinum-resistant ovarian cancer, antibody-drug conjugate, corneal toxicity, peripheral neuropathy, folate receptor alpha, TriNetX, real-world evidence, propensity score matching, drug safety, ovarian cancer, electronic health records