ChatGPT accurately performs genetic counseling for gynecologic cancers

OBJECTIVE

Artificial Intelligence (AI) systems such as ChatGPT can take medical examinations and counsel patients regarding medical diagnosis. We aim to quantify the accuracy of the ChatGPT V3.4 in answering commonly asked questions pertaining to genetic testing and counseling for gynecologic cancers.

METHODS

Forty questions were formulated in conjunction with gynecologic oncologists and adapted from professional society guidelines and ChatGPT version 3.5 was queried, the version that is readily available to the public. The two categories of questions were genetic counseling guidelines and questions pertaining to specific genetic disorders. The answers were scored by two attending Gynecologic Oncologists according to the following scale: 1) correct and comprehensive, 2) correct but not comprehensive, 3) some correct, some incorrect, and 4) completely incorrect. Scoring discrepancies were resolved by additional third reviewer. The proportion of responses earning each score were calculated overall and within each question category.

RESULTS

ChatGPT provided correct and comprehensive answers to 33/40 (82.5%) questions, correct but not comprehensive answers to 6/40 (15%) questions, partially incorrect answers to 1/40 (2.5%) questions, and completely incorrect answers to 0/40 (0%) questions. The genetic counseling category of questions had the highest proportion of answers that were both correct and comprehensive with ChatGPT answering all 20/20 questions with 100% accuracy and were comprehensive in responses. ChatGPT performed equally in the specific genetic disorders category, with 88.2% (15/17) and 66.6% (2/3) correct and comprehensive answers to questions pertaining to hereditary breast and ovarian cancer and Lynch syndrome questions respectively.

CONCLUSION

ChatGPT accurately answers questions about genetic syndromes, genetic testing, and counseling in majority of the studied questions. These data suggest this powerful tool can be utilized as a patient resource for genetic counseling questions, though more data input from gynecologic oncologists would be needed to educate patients on genetic syndromes.

Bridging Communication Gaps Between Radiologists, Referring Physicians, and Patients Through Standardized Structured Cancer Imaging Reporting: The Experience with Female Pelvic MRI Assessment Using O-RADS and a Simulated Cohort Patient Group

RATIONALE AND OBJECTIVES

This study aimed to evaluate whether implementing structured reporting based on Ovarian-Adnexal Reporting and Data System (O-RADS) magnetic resonance imaging (MRI) in women with sonographically indeterminate adnexal masses improves communication between radiologists, referrers, and patients/caregivers and enhances diagnostic performance for determining adnexal malignancy.

MATERIALS AND METHODS

We retrospectively analyzed prospectively issued MRI reports in 2019-2022 performed for characterizing adnexal masses before and after implementing O-RADS MRI; 56 patients/caregivers and nine gynecologic oncologists ("referrers") were surveyed about report interpretability/clarity/satisfaction; responses for pre- and post-implementation reports were compared using Fisher's exact and Chi-squared tests. Diagnostic performance was assessed using receiver operating characteristic curves.

RESULTS

A total of 123 reports from before and 119 reports from after O-RADS MRI implementation were included. Survey response rates were 35.7% (20/56) for patients/caregivers and 66.7% (6/9) for referrers. For patients/caregivers, O-RADS MRI reports were clearer (p < 0.001) and more satisfactory (p < 0.001) than unstructured reports, but interpretability did not differ significantly (p = 0.14), as 28.0% (28/100) of postimplementation and 38.0% (38/100) of preimplementation reports were considered difficult to interpret. For referrers, O-RADS MRI reports were clearer, more satisfactory, and easier to interpret (p < 0.001); only 1.3% (1/77) were considered difficult to interpret. For differentiating benign from malignant adnexal lesions, O-RADS MRI showed area under the curve of 0.92 (95% confidence interval [CI], 0.85-0.99), sensitivity of 0.81 (95% CI, 0.58-0.95), and specificity of 0.91 (95% CI, 0.83-0.96). Diagnostic performance of reports before implementation could not be calculated due to many different phrases used to describe the likelihood of malignancy.

CONCLUSION

Implementing standardized structured reporting using O-RADS MRI for characterizing adnexal masses improved clarity and satisfaction for patients/caregivers and referrers. Interpretability improved for referrers but remained limited for patients/caregivers.

Let's chat about cervical cancer: Assessing the accuracy of ChatGPT responses to cervical cancer questions

OBJECTIVE

To quantify the accuracy of ChatGPT in answering commonly asked questions pertaining to cervical cancer prevention, diagnosis, treatment, and survivorship/quality-of-life (QOL).

METHODS

ChatGPT was queried with 64 questions adapted from professional society websites and the authors' clinical experiences. The answers were scored by two attending Gynecologic Oncologists according to the following scale: 1) correct and comprehensive, 2) correct but not comprehensive, 3) some correct, some incorrect, and 4) completely incorrect. Scoring discrepancies were resolved by additional reviewers as needed. The proportion of responses earning each score were calculated overall and within each question category.

RESULTS

ChatGPT provided correct and comprehensive answers to 34 (53.1%) questions, correct but not comprehensive answers to 19 (29.7%) questions, partially incorrect answers to 10 (15.6%) questions, and completely incorrect answers to 1 (1.6%) question. Prevention and survivorship/QOL had the highest proportion of "correct" scores (scores of 1 or 2) at 22/24 (91.7%) and 15/16 (93.8%), respectively. ChatGPT performed less well in the treatment category, with 15/21 (71.4%) correct scores. It performed the worst in the diagnosis category with only 1/3 (33.3%) correct scores.

CONCLUSION

ChatGPT accurately answers questions about cervical cancer prevention, survivorship, and QOL. It performs less accurately for cervical cancer diagnosis and treatment. Further development of this immensely popular large language model should include physician input before it can be utilized as a tool for Gynecologists or recommended as a patient resource for information on cervical cancer diagnosis and treatment.

Sentinel lymph node biopsy alone compared to systematic lymphadenectomy in patients with uterine carcinosarcoma

OBJECTIVE

To assess survival among patients diagnosed with uterine carcinosarcoma (CS) who underwent sentinel lymph node (SLN) biopsy alone vs. systematic lymph node dissection (LND).

METHODS

We identified newly diagnosed CS patients who underwent primary surgical management from January 1996-December 2019. The SLN cohort underwent SLN biopsy alone with bilateral SLNs identified. The systematic LND cohort did not undergo SLN biopsy.

RESULTS

Ninety-nine patients underwent SLN biopsy, and 100 patients underwent systematic LND. There was no difference by age, stage, body mass index, myoinvasion (<50%, ≥50%), lymphovascular space invasion, or positive washings. Eighty-five SLN (85.9%) and 15 LND (15%) underwent minimally invasive surgery (P < 0.001). The median total node count was four (range, 1-13) for SLN and 19 (range, 2-50) for LND (P < 0.001). Nodal metastasis occurred in 23 (23.2%) SLN and in 22 (22%) LND (P = 0.4). Postoperative therapy was administered to 85 (85.9%) SLN and 71 (71%) LND (P = 0.02). Median follow-up was 33 months (range, 1-205) for SLN and 55.3 months (range, 1-269) for LND (P = 0.001). The three-year progression-free survival (PFS) was 62.9% (SE 5.2%) for SLN and 52.3% (SE 5.3%) for LND (P = 0.13). The three-year overall survival (OS) was 72.1% (SE 5.1%) for SLN and 71.6% (SE 4.6%) for LND (P = 0.68). An isolated nodal recurrence occurred in two (2%) SLN and four (4%) LND (P = 0.26).

CONCLUSIONS

There is no difference in PFS or OS among CS patients who undergo SLN biopsy vs. systematic LND. SLN biopsy detects nodal metastasis without compromising oncologic outcomes.

Patient-reported benefit from proposed interventions to reduce financial hardship during cancer treatment

7080

Background: Awareness of cancer patients’ financial toxicity (FT) has increased substantially over the past decade; however, interventions to minimize financial burden remain underdeveloped and understudied. This survey-based study explores patient beliefs on which potential mitigating strategies could improve their financial hardship during cancer treatment. Methods: Interviewer-administered surveys were conducted with consecutive patients in an outpatient, urban, private academic Gynecologic Cancer clinic waiting room for 2 weeks in August 2019. The survey items included patient demographics, disease characteristics, the Comprehensive Score for Financial Toxicity (COST) tool (validated measure of FT with score 0-44; lower scores indicate worse FT), assessment of cost-coping strategies, and patient-reported anticipated benefit from described potential interventions (items that were feasible and relevant to implement in clinic). Results: Of 101 patients who initiated the survey, 87 (86%) completed it and were included in this analysis. The median age was 66 (range, 32-87). Thirty-eight patients (44%) had ovarian, 29 (33%) uterine, 5 (6%) cervical, and 15 (17%) an “other” gynecologic cancer. The median COST score was 32 (range, 6-44). Twenty-nine patients (33%) had COST scores ≤25 and 16 (18%) had COST scores ≤18. The most frequent cost-coping strategy reported was reducing leisure activities (n = 36, 41%) and using savings to pay for medical bills (n = 34, 39%). Six patients (7%) reported not taking a prescribed medication in the past 12 months due to the inability to pay and 0 reported skipping a recommended imaging study. When it came to interventions patients anticipated would improve their current financial hardships, 34 (39%) indicated access to transportation assistance to and from appointments, 31 (36%) said “knowing up front how much I’m going to have to pay for my healthcare”, 29 (33%) indicated “minimizing wait time associated with appointments, which keeps me away from work”, and 22 (25%) indicated “access to free food during/around appointments and treatments”. Only 26 (30%) noted they were not experiencing financial hardship. Conclusions: For an outpatient population of gynecologic cancer patients, several focused, feasible interventions could be implemented to potentially decrease patient FT. Our study can help health care providers in the design of interventions to create meaningful improvements in patient financial burden. Next steps should assess the impact of targeted interventions on patient outcomes.

Lower extremity lymphedema in patients with gynecologic malignancies

Lower extremity lymphedema is a chronic, often irreversible condition that affects many patients treated for gynecologic malignancies, with published rates as high as 70% in select populations. It has consistently been shown to affect multiple quality of life metrics. This review focuses on the pathophysiology, incidence, trends, and risk factors associated with lower extremity lymphedema secondary to the treatment of cervical, endometrial, ovarian, and vulvar cancers in the era of sentinel lymph node mapping. We review traditional and contemporary approaches to diagnosis and staging, and discuss new technologies and imaging modalities. Finally, we review the data-based treatment of lower extremity lymphedema and discuss experimental treatments currently being developed. This review highlights the need for more prospective studies and objective metrics, so that we may better evaluate and serve these patients.

Patient preference for receiving information about long-term symptoms

e21681

Background: Maintaining quality of life with minimal symptoms is an important goal of cancer treatment. We surveyed patients undergoing cancer treatment about their preferences for information about the long-term symptom burden associated with their current treatment. Methods: Patients undergoing active cancer treatment were enrolled at two large cancer centers from 1/2014 to 7/2015. At the time of enrollment patients reported their household income, employment, performance status and their agreement with the statement, “It would be useful to know the percentage of patients having persistent symptoms more than a year after getting the same treatment I am getting.” Treatment intention (curative vs. palliative) was reported by the treating oncologist. Results: 196 patients completed the survey. The mean age was 55.7 (range 20-80); 54% were female, 69% were white, and 60% were receiving palliative treatment. The majority (83%) of patients either strongly agreed or somewhat agreed that it would be helpful to have information about future symptoms. In univariable analyses, this preference did not significantly differ based on patient-reported performance status, sex, age, treatment intent, race, or household income. Using a more stringent criterion to dichotomize the outcome, 59% of participants strongly agreed that it would be helpful to have information about future symptoms. Patients undergoing treatment with curative intent were significantly more likely to strongly agree (72% vs. 51%, p = 0.0035). In a multivariable analysis adjusting for age, sex, income level, employment status, and patient-reported performance status, participants undergoing curative versus palliative treatment had 3.1 times the odds (95% CI 1.47-6.65) of strongly agreeing. Conclusions: The majority of patients undergoing cancer treatment wanted to know what symptoms to expect in the future, and this was independent of patient demographics. Patients undergoing treatment with curative intent are significantly more likely to want this information than patients undergoing palliative treatment. Symptom data collected through research studies and registries should be summarized by treatment and other key predictors and made available to patients.

A multicenter analysis of patient reported risk factors for not working during cancer therapy

6523

Background: Treatment for most types of cancer results in work disruptions. The aim of this study is to identify risk factors for not working during cancer therapy using a patient-reported outcome tool. Methods: Patients of all cancer types, who were undergoing treatment with curative or palliative intent, were enrolled in a survey-based study at Memorial Sloan Kettering Cancer Center and the North Carolina Cancer Hospital from 1/2014 to 7/2015. All patients working full- or part- time at the time of cancer diagnosis were included in this analysis. Patient reported outcomes were collected using a survey tool at any time during active cancer treatment. The primary outcome of this analysis was self-reported work status at the time of survey completion. Results: Of 119 patients who were working before their cancer diagnosis, 68% were working at the time of survey completion. The mean age was 52.9 (range 20-80). Younger age was associated with working: 82% of patients age < 40 were working during treatment vs. 31% of patients age > 65 (p=0.01). The number of days patients reported being completely unable to work in the past month was significantly associated with work status; only 26% of patients who missed > 10 days continued to work compared to 69% who missed ≤ 10 days (p<0.01). Patients who report having a flexible work schedule were 23% more likely to continue to work during treatment (69% vs. 46%, p=0.01). In a multivariable model controlling for having an employer contingent health plan, and household income > 200% of the federal poverty level (fpl), only age, having a flexible work schedule, and inability to work > 10 days were independently associated with work status during treatment (Table). Conclusions: Having a flexible work schedule allowed for more work interruptions without job loss. Patients who are likely to miss more than 10 days of work and who have inflexible work schedules are at high risk of unemployment and should be targeted for employment retention initiatives. [Table: see text]

Genetic Basis for PD-L1 Expression in Squamous Cell Carcinomas of the Cervix and Vulva

IMPORTANCE

Patients with squamous cell carcinoma (SCC) of the cervix or vulva have limited therapeutic options, and the potential for immunotherapy for this population has not been evaluated. Recent trials suggest that tumors with a genetic basis for PD-1 (programmed cell death protein 1) ligand expression are highly sensitive to therapeutic antibodies targeting PD-1.

OBJECTIVE

To determine the genetic status of CD274 (encoding PD-L1 [programmed cell death 1 ligand 1]) and PDCD1LG2 (encoding PD-L2 [programmed cell death 1 ligand 2]) in SCCs of the cervix and vulva and to correlate the findings with PD-L1 protein expression.

DESIGN, SETTING, AND PARTICIPANTS

We performed fluorescence in situ hybridization (FISH) using probes targeting CD274, PDCD1LG2, and the centromeric portion of chromosome 9, and immunohistochemistry (IHC) using an antibody recognizing PD-L1 on formalin-fixed, paraffin-embedded (FFPE) biopsy specimens from 48 cervical SCCs and 23 vulvar SCCs.

MAIN OUTCOMES AND MEASURES

Tumors were categorized according to the genetic abnormality in CD274 and PDCD1LG2 (coamplification > cogain > polysomy > disomy) as detected by FISH, and evaluated on a semiquantitative scale (modified H score, the product of the percentage of tumor cells with positive staining and the maximum intensity of positive staining) for PD-L1 protein expression as detected by IHC.

RESULTS

Overall, 71 samples of FFPE tissue from cases of cervical SCCs (n = 48) and vulvar SCCs (n = 23) were retrieved from the archives of Brigham and Women's Hospital and included in this study. We observed cogain or coamplification of CD274 and PDCD1LG2 in 32 of 48 cervical SCCs (67%) and 10 of 23 vulvar SCCs (43%). Median PD-L1 protein expression was highest among tumors with CD274 and PDCD1LG2 coamplification and lowest among tumors with disomy.

CONCLUSIONS AND RELEVANCE

Recurrent copy number gain of the genes encoding the PD-1 ligands provides a genetic basis for PD-L1 expression in a subset of cervical and vulvar SCCs and identifies a class of patients that are rational candidates for therapies targeting PD-1.