Diagnóstico de hirsutismo por meio de análise de imagens, autoavaliação e escore de Ferriman-Gallwey simplificado

Abstract

Hirsutism, defined as hair growth in women with a male distribution pattern, affects 5-10% of the female population. It represents a clinical sign of hyperandrogenism and is present in up to 80% of patients with Polycystic Ovary Syndrome (PCOS). Although there are ethnic variations, the diagnosis of hirsutism is based on the modified Ferriman-Gallwey score (mFG) ≥ 6 (Endocrine Society) and ≥ 4 (European Society of Human Reproduction and Embryology), endorsed by the latest guideline joint effort of these societies, (Recommendations from the 2023 international evidence-based guideline for the assessment and management of polycystic ovary syndrome), published in the second half of 2023. Important limitations in the use of this score in clinical practice have already been described: the need for a trained evaluator of the procedure, use of cosmetics (hair whitening) that make clinical assessment difficult, low inter-observer agreement, discrepancy between the severity of hirsutism and the patient's perception and the need for the patient to be undressed and not shaved for correct assessment. Therefore, the creation of a reproducible and easy-to-perform method represents an advance in the clinical diagnosis of hirsutism. The results presented in the articles attached to this dissertation are derived from a cross-sectional study with women aged 18-50 years treated at the Gynecology and Obstetrics and Endocrinology and Metabology outpatient clinics of the Hospital das Clínicas of UFMG between July 2021 and September 2023. We collected information about a diagnosis, medications in use, clinical and menstrual history, as well as complaints about excess hair, and the self-reported Ferriman-Gallwey score. Trained examiners obtained the mF-G (gold standard), evaluated and took photos using a cell phone camera of the nine regions that make up the score to later obtain the Ferriman-Gallwey score per image (mFG-image). Subsequently, the mFG-image, the self-reported Ferriman- Gallwey score, and the mini-score (simplified score obtained from the analysis of the upper abdomen, lower abdomen, and chin) were compared to the gold standard. In the work that compared the mF-G-image with the gold standard, 70 women were included. Among the examiners who analyzed the pictures, inter-observer agreement was 81.4% (Fleiss Kappa), with a Kappa Index of 0.75. For an mFG score ≥ 6, agreement was 77% and the Kappa Index was 0.62 (moderate). Bland–Altman analysis established an agreement of 0.89 (95% CI [0.83, 0.92]) between in-person and image-based methods for scoring mFG. In the mini-score assessment (involving the use of 3 areas of the modified Ferriman-Gallwey scale (chin, lower abdomen, and upper abdomen) to establish hirsutism with a cutoff point > 3) 188 participants were included. The accuracy, positive predictive value (PPV) and negative predictive value (NPV) of the mini-score were, respectively, 94.7%, 76.9%, and 99.1% considering a mF-G cutoff ≥ 6 and 89.9%, 86.2%, and 91.0%, considering the mF-G cutoff ≥ 4 for the diagnosis of hirsutism. In the third study, evaluating the usefulness of the self-reported Ferriman-Gallwey score, 188 participants were evaluated and sensitivity, specificity and NPV were, respectively, 95.5% (77.1-99.%), 55.4% (45.7 -64.8%), NPV 98.4% using mF-G ≥ 6 and 94.9% (82.7) as a cutoff for hirsutism -99.4%), 37.4% (45.7 -64.8%) and NPV: 94.4% for mF-G cutoff ≥ 4. In addition to these results, a publication of a narrative review on the importance and difficulties of assessing hirsutism was attached in women, highlighting the challenge of identifying women who have some suffering related to increased hair even without a very high Ferriman score. Taken together, the results of this dissertation confirm that 1) the Ferriman-Gallwey auto-score is a low specificity tool, but with the possibility of being used to rule out hirsutism due to its high NPV; 2) the use of images (such as those obtained by a cell phone) offers excellent reproducibility and can be used by trained evaluators to diagnose hirsutism in a non-face-toface manner; 3) the use of the miniscore in a Latin American population may be another useful alternative for establishing hirsutism, being performed more quickly with exposure of only the chin, lower and upper abdomen. Future perspectives involve the study of diagnosing hirsutism using artificial intelligence using imaging devices.