br ovarian cancer with lower confidence interval boundaries
+ ovarian cancer with lower 95% confidence interval boundaries of
81.19–87.75 depending on the statistical model used to analyze the data. The GLMM analysis that accounts for possible correlation of lesions within the patient estimated the higher sensitivity of 97.97%. Even in the unselected (FRa+ and FRa− combined) population, OTL38 achieved sen-sitivity and PPV comparable to that shown in the FRa+ specific patients. This is relevant because FRa status is not typically known pre-operatively, and this suggests that neither unknown FRa status nor FRa− status would contraindicate the use of OTL38 in surgery. Also, in Dynasore to the phase I study, this expanded cohort revealed an additional 48% of pa-
tients who had at least one lesion identified by imaging alone, and 40% of the patients with miliary disease having 4 or more that would have been left behind without the use of the investigational agent. Because removal of normal tissues was not built into the study de-
sign, a true specificity of OTL38 could not be calculated. However, 29 false positive lesions were detected in 11 patients by each of the 4 inves- tigators (Table 4). Most of the lesions classified as false positives expressed FR-beta and were located primarily in the lymph nodes, though 2 true positive lymph nodes were misclassified by the on-site pathologist. Falsely positive lymph nodes were also observed in the pre- ceding phase I study of OTL38 in ovarian cancer . These findings should be considered in the context of the prospective, randomized Lymphadenectomy in Ovarian Neoplasms (LION) trial, which reported excess morbidity and mortality from the removal of clinically negative lymph nodes, 56% of which were microscopically positive, with no gain in survival time . Therefore this technology has the potential to increase surgical risk if all NIR+ nodes are removed. Further studies are needed to confirm and further characterize falsely positive lymph nodes. When analyzed in the phase 1 trial, falsely positive nodes with FRb expression were a marker for the presence of tumor-activated mac-rophages that have potential immunotherapeutic relevance [15,16].
Toxicity in this study was consistent with that observed in the pre-ceding phase I study  and was not dose-limiting. When considering all toxicity irrespective of study drug attribution, all patients experi-enced at least one adverse event. The most common AE was procedural pain and occurred independent of the study drug. The most common drug-related AEs/TEAEs were gastrointestinal related nausea, vomiting, and abdominal pain. While these AEs are considered probably related to the study drug, it is possible that concurrent medications given in the pre-operative setting such as intravenous narcotics could also cause these effects. However, a dose-response increase in these AEs was noted in the phase I study, increasing the likelihood that these are truly drug related. Given the typical disease distribution of ovarian can-cer, miliary disease deposits involving the gastrointestinal tract, there might be a biologic rationale for this consistently observed effect. Re-gardless, drug-attributable toxicity with this agent is mild and resolves within 24 h of study drug receipt.
The development of an imaging adjunct to ovarian cancer surgery that can be applied “real-time” during surgery has many potential appli-cations. First, such a technique could improve the accuracy of cancer staging that is performed in the setting of an isolated ovarian mass. The standard staging operation relies on systematic removal of lymph nodes and omentum with peritoneal biopsies of tissues that appear nor-mal and cancer free. In this situation, targeted imaging could allow de-tection of metastatic lesions that are not visible under white light and,