Editors: Rouzan Karabakhtsian, MD, PhD, professor of pathology and director of the Women’s Health Pathology Fellowship, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY; S. Emily Bachert, MD, associate pathologist, Brigham and Women’s Hospital, Boston; Amarpreet Bhalla, MD, assistant professor of pathology, Albert Einstein College of Medicine, Montefiore Medical Center; Divya Sharma, MD, associate professor, Department of Pathology and Laboratory Medicine, University of Cincinnati Medical Center; and Paula Toro, MD, pathology resident, Cleveland Clinic.
Validation study of newly proposed refined diagnostic criteria for malignant phyllodes tumor
January 2025—The World Health Organization’s diagnostic criteria for malignant phyllodes tumor may miss a significant number of such tumors that have metastatic potential, according to a study conducted by the authors. Therefore, the authors proposed new refined diagnostic criteria for malignant phyllodes tumor (MPT) and conducted a study to validate the refined criteria. Their validation study included 136 borderline phyllodes tumors (BoPTs) and MPT cases that were not included in the initial study. The authors evaluated tumor classifications based on the refined criteria and World Health Organization (WHO) criteria. The revised criteria for MPT were either stromal overgrowth and one or more other features, such as marked stromal cellularity, marked stromal cytologic atypia, or at least 10 mitoses per 10 high-power fields (10 mitoses/10 HPF), or it was absence of stromal overgrowth and one or more other features, such as marked stromal cytologic atypia, at least 10 mitoses/10 HPF, or permeative border. The WHO criteria require all five morphologic features—stromal overgrowth, permeative border, marked stromal cellularity, marked stromal cytologic atypia, and 10 or more mitoses/10 HPF—for an MPT diagnosis. Using the refined criteria, none of the 61 BoPTs developed metastases and 40 percent of the 75 MPTs developed metastases. Local recurrence was seen in 11.5 percent of BoPTs and 25.3 percent of MPTs. Using the WHO criteria, 9.6 percent of the 94 BoPTs developed metastases and 50 percent of the 42 MPTs developed metastases. In addition, 14.9 percent of the BoPTs and 28.6 percent of the MPTs had local recurrence. Nine (30 percent) of the 30 tumors that developed distant metastases were diagnosed as BoPTs using the WHO criteria. When the authors combined the 75 MPTs from this validation cohort with the 65 MPT cases from the published data using the refined criteria, 50 (35.7 percent) of the 140 MPTs developed metastases. In the univariate analysis with log-rank test, stromal overgrowth, marked stromal cellularity, marked stromal cytologic atypia, 10 or more mitoses/10 HPF, presence of heterologous components other than liposarcomatous component, and presence of stromal necrosis were significantly associated with risk of metastasis (all, P<.05). In multivariate analysis with Cox proportional hazard regression, stromal overgrowth and marked stromal cellularity were significantly associated with metastasis (both, P<.001). The authors determined that the refined criteria are superior to the WHO criteria in predicting the clinical outcomes of BoPTs and MPTs. Using the refined criteria, 35.7 percent of 140 patients with MPT developed metastases, whereas none of the patients with BoPT developed metastases. Patients with MPT have a high metastatic rate and may benefit from systemic chemotherapy or targeted therapies. In contrast, patients with BoPTs may be managed with complete local excision alone without chemotherapy.
Li X, Nguyen TTA, Zhang J, et al. Validation study of the newly proposed refined diagnostic criteria for malignant phyllodes tumor with 136 borderline and malignant phyllodes tumor cases. Am J Surg Pathol. 2024;48(9):1146–1153.
Correspondence: Dr. Xiaoxian Li at xli40@emory.edu
A prospective study using AI to improve dermoscopic melanoma diagnosis in patient care
Early detection of melanoma, a potentially lethal type of skin cancer that is highly prevalent worldwide, improves patient prognosis. Recent retrospective studies have shown artificial intelligence (AI) to be helpful in enhancing melanoma detection. However, few prospective studies have been conducted to confirm these promising results, and the studies have been limited by small sample sizes, homogenous data sets, and lack of inclusion of rare melanoma subtypes, preventing a fair and thorough evaluation of AI and its generalizability. To address these limitations, the authors assessed All Data are Ext (ADAE), an established open-source ensemble algorithm for detecting melanoma, by comparing its diagnostic accuracy to that of dermatologists on a prospectively collected, external, heterogeneous test set comprising eight university hospitals, four distinct camera setups, rare melanoma subtypes, and special anatomical sites. They advanced the algorithm with real test-time augmentation—that is, providing photographs of lesions taken from multiple angles and averaging the predictions—and evaluated its generalization capabilities. Overall, AI showed higher balanced accuracy than dermatologists (0.798, 95 percent confidence interval [CI], 0.779–0.814 versus 0.781, 95 percent CI, 0.760–0.802; p=4.0e−145). It obtained higher sensitivity (0.921, 95 percent CI, 0.900–0.942 versus 0.734, 95 percent CI, 0.701–0.770; p=3.3e−165) at the cost of lower specificity (0.673, 95 percent CI, 0.641–0.702 versus 0.828, 95 percent CI, 0.804–0.852; p=3.3e−165). The authors concluded that because the algorithm exhibited a significant performance advantage in their heterogeneous data set that exclusively comprised melanoma-suspicious lesions, AI potentially can support dermatologists, particularly in diagnosing challenging cases.
Heinlein L, Maron RC, Hekler A, et al. Prospective multicenter study using artificial intelligence to improve dermoscopic melanoma diagnosis in patient care. Commun Med. 2024. doi.org/10.1038/s43856-024-00598-5
Correspondence: Dr. Titus J. Brinker at titus.brinker@dkfz.de
Interobserver agreement in interpretation of anal cytology
Anal cytology is used for anal cancer screening in high-risk populations. In addition to accuracy, the reproducibility of the interpretation is of key importance. The authors conducted a study to evaluate interobserver agreement in anal cytology interpretation. For the study, two cytopathologists who had at least 10 years of experience in cervicovaginal cytology independently evaluated the liquid-based cytologic slides for HIV-negative men who have sex with men (MSM). Cases with a discordant interpretation were reviewed and a consensus reached. Human papillomavirus (HPV) genotyping was performed using a proprietary HPV genotyping test. Unweighted and weighted Cohen kappa and 95 percent confidence interval (CI) values were calculated. A total of 713 slides that were adequate for interpretation were evaluated (median age of MSM, 33 years). An HPV test was performed on 620 samples (86.9 percent). Considering a dichotomous interpretation (negative for intraepithelial lesion or malignancy versus atypical squamous cells of undetermined significance or worse), the crude agreement between the two readers was 93.3 percent (κ=0.82; 95 percent CI, 0.77–0.87). Once a consensus for discordant cases was reached, the negative for intraepithelial lesion or malignancy category (511 of 528 samples, 96.8 percent) had the best agreement, and the atypical squamous cells of undetermined significance category showed the lowest agreement (90 of 117 samples, 76.9 percent). When taking into consideration the individual cytologic categories, the overall agreement was 92.1 percent (κ=0.85; 95 percent CI, 0.81–0.89). The discordant interpretations were not associated with high-risk HPV infection, HPV16 infection, or age of MSM. The results of this study substantiate the use of anal cytology as a cancer screening strategy for HIV-negative MSM.
Benevolo M, Rollo F, Latini A, et al. Interobserver agreement in the interpretation of anal cytology. Cancer Cytopathol. 2024;132:419–424.
Correspondence: Dr. Francesca Rollo at francesca.rollo@ifo.it
Comparative analysis of hepatocellular neoplasm, not otherwise specified, and hepatoblastoma
Accurate diagnosis and treatment of hepatocellular neoplasm, not otherwise specified (HCN-NOS), poses significant challenges. The authors conducted a study to investigate the clinicopathologic and genomic similarities and differences between HCN-NOS and hepatoblastoma to guide diagnostic and treatment strategies. The clinicopathologic characteristics of 16 patients with HCN-NOS and 23 patients with hepatoblastoma were compared. The molecular analysis employed the OncoKids DNA- and RNA-based next-generation sequencing panel, chromosomal microarray, and targeted Sanger sequencing analyses of CTNNB1 and TERT promoter mutations. The investigators found that patients with HCN-NOS were older (P<.001) and more frequently classified as high risk (P<.01), yet they showed no significant differences in α-fetoprotein levels or survival outcomes when compared with those who had hepatoblastoma. HCN-NOS and hepatoblastoma had a comparable frequency of sequence variants, and CTNNB1 mutations were predominant in both groups. Notably, TERT promoter mutations (37.5 percent) and rare clinically significant variants (BRAF, NRAS, and KMT2D) were exclusive to HCN-NOS. The latter demonstrated a higher prevalence of gains in 1q, encompassing the MDM4 locus (17 of 17 versus 11 of 24; P<.001), and loss/loss of heterozygosity (LOH) of 1p (11 of 17 versus six of 24; P<.05) and chromosome 11 (seven of 17 versus one of 24; P<.01) when compared with hepatoblastoma. Furthermore, the recurrent loss/LOH of chromosomes 3, 4p, 9, 15q, and Y was observed only in HCN-NOS. However, no significant differences in gains of chromosomes 2, 8, and 20 or in loss/LOH of 4q and 11p were noted between the two groups. Notably, no clinically significant gene fusions were detected in either group. The authors concluded that this study reveals that HCN-NOS exhibits high-risk clinicopathologic features and greater structural complexity than hepatoblastoma. However, both groups exhibit comparable α-fetoprotein levels at diagnosis, CTNNB1 mutation rates, and survival outcomes when subjected to aggressive treatment. These findings have the potential to enhance diagnostic accuracy and inform more effective treatments for HCN-NOS.
Zhou S, Sarabia SF, Estrine D, et al. Comparative clinicopathologic and genomic analysis of hepatocellular neoplasm, not otherwise specified, and hepatoblastoma. Mod Pathol. 2024. doi.org/10.1016/j.modpat.2023.100385
Correspondence: Dr. Shengmei Zhou at szhou@chla.usc.edu