Multicenter automatic detection of invasive carcinoma on breast whole slide images.

Breast cancer is one of the most prevalent cancers worldwide and pathologists are closely involved in establishing a diagnosis. Tools to assist in making a diagnosis are required to manage the increasing workload. In this context, artificial intelligence (AI) and deep-learning based tools may be used in daily pathology practice. However, it is challenging to develop fast and reliable algorithms that can be trusted by practitioners, whatever the medical center. We describe a patch-based algorithm that incorporates a convolutional neural network to detect and locate invasive carcinoma on breast whole-slide images. The network was trained on a dataset extracted from a reference acquisition center. We then performed a calibration step based on transfer learning to maintain the performance when translating on a new target acquisition center by using a limited amount of additional training data. Performance was evaluated using classical binary measures (accuracy, recall, precision) for both centers (referred to as "test reference dataset" and "test target dataset") and at two levels: patch and slide level. At patch level, accuracy, recall, and precision of the model on the reference and target test sets were 92.1% and 96.3%, 95% and 87.8%, and 73.9% and 70.6%, respectively. At slide level, accuracy, recall, and precision were 97.6% and 92.0%, 90.9% and 100%, and 100% and 70.8% for test sets 1 and 2, respectively. The high performance of the algorithm at both centers shows that the calibration process is efficient. This is performed using limited training data from the new target acquisition center and requires that the model is trained beforehand on a large database from a reference center. This methodology allows the implementation of AI diagnostic tools to help in routine pathology practice.

Cutaneous vasculitis occurring in the setting of systemic lupus erythematosus: a multicentre cohort study.

OBJECTIVES: To describe the clinical and pathological features of biopsy-proven cutaneous vasculitis (CV) associated with SLE, focusing on diagnosis classification and impact on overall SLE activity. METHODS: Retrospective multicentric cohort study including SLE patients with biopsy-proven CV identified by (i) data from pathology departments of three university hospitals and (ii) a national call for cases. SLE was defined according to 1997 revised ACR and/or 2019 ACR/EULAR criteria. CV diagnosis was confirmed histologically and classified by using the dermatological addendum of the Chapel Hill classification. SLE activity and flare severity at the time of CV diagnosis were assessed independently of vasculitis items with the SELENA-SLEDAI and SELENA-SLEDAI Flare Index. RESULTS: Overall, 39 patients were included; 35 (90%) were female. Cutaneous manifestations included mostly palpable purpura (n = 21; 54%) and urticarial lesions (n = 18; 46%); lower limbs were the most common location (n = 33; 85%). Eleven (28%) patients exhibited extracutaneous vasculitis. A higher prevalence of Sjögren's syndrome (51%) was found compared with SLE patients without CV from the French referral centre group (12%, P < 0.0001) and the Swiss SLE Cohort (11%, P < 0.0001). CV was mostly classified as urticarial vasculitis (n = 14, 36%) and cryoglobulinaemia (n = 13, 33%). Only 2 (5%) patients had no other cause than SLE to explain the CV. Sixty-one percent of patients had inactive SLE. CONCLUSION: SLE-related vasculitis seems very rare and other causes of vasculitis should be ruled out before considering this diagnosis. Moreover, in more than half of patients, CV was not associated with another sign of active SLE.

Myeloid Clonal Infiltrate Identified With Next-Generation Sequencing in Skin Lesions Associated With Myelodysplastic Syndromes and Chronic Myelomonocytic Leukemia: A Case Series.

Background: Myelodysplastic syndromes (MDS) and chronic myelomonocytic leukemia (CMML) are associated with cutaneous manifestations. Next-generation sequencing (NGS) is a tool capable of identifying clonal myeloid cells in the skin infiltrate and thus better characterize the link between hematological diseases and skin lesions. Objective: To assess whether skin lesions of MDS/CMML are clonally related to blood or bone marrow cells using NGS. Methods: Comparisons of blood or bone marrow and skin samples NGS findings from patients presenting with MDS/CMML and skin lesions in three French hospitals. Results: Among the 14 patients recruited, 12 patients (86%) had mutations in the skin lesions biopsied, 12 patients (86%) had a globally similar mutational profile between blood/bone marrow and skin, and 10 patients (71%) had mutations with a high variant allele frequency (>10%) found in the myeloid skin infiltrate. Mutations in TET2 and DNMT3A, both in four patients, were the most frequent. Two patients harbored a UBA1 mutation on hematopoietic samples. Limitations: Limited number of patients and retrospective collection of the data. Blood and skin sampling were not performed at the exact same time point for two patients. Conclusion: Skin lesions in the setting of MDS/CMML are characterized by a clonal myeloid infiltrate in most cases.

Low risk of invasive lobular carcinoma of the breast in carriers of BRCA1 (hereditary breast and ovarian cancer) and TP53 (Li-Fraumeni syndrome) germline mutations.

BACKGROUND: Invasive lobular carcinoma (ILC) of the breast has epidemiological, molecular and clinical specificities, and should likely be considered a unique entity. As for genetic susceptibility, CDH1 germline mutations predispose exclusively to ILC. Data are however scarce regarding ILC in women with BRCA1/2 (Hereditary Breast and Ovarian Cancer) and TP53 (Li-Fraumeni syndrome) germline mutations. METHODS: We included all breast cancers from female patients tested at our institute between 1992 and 2016 (n = 3469) for which pathology data were available. ILC proportion comparison according to mutational status was performed by a chi-squared test. The impact of susceptibility genes on ILC proportion was investigated by univariate logistic regression with wild-type patients as reference. RESULTS AND DISCUSSION: There were 265 (7.64%) ILC: 2/342 (0.58%) in BRCA1 patients, 24/238 (10%) in BRCA2 patients, 1/57 (1.75%) in TP53 patients and 238/2832 (8.4%) in non-carriers. The majority of breast cancers in all groups were invasive ductal and ductal in situ carcinomas. The difference in ILC proportion was highly significant (P < 0.001). Compared to wild-type patients, BRCA1 was associated with a lower ILC proportion (OR 0.064 [95% CI 0.016;0.259], P < 0.0001). BRCA2 OR was 1.222 [95%CI 0.785;1.902] (P = 0.374), TP53 OR was 0.195 [95%CI 0.027;1.412] (P = 0.105). ILC are therefore underrepresented in BRCA1 and TP53 mutation carriers. Formal significance (P = 0.05) was not reached for TP53, but statistical power was only 38%. Based on ILC incidence in the general population, we make the hypothesis that BRCA1 and TP53 do not predispose to ILC, as the few occurrences of ILC in mutation carriers could be attributed to chance and not to germline mutations. Our observations will be useful to clinical cancer geneticists managing patients with ILC, as a BRCA1 or TP53 mutation in these patients would be unlikely. Genetic counseling should be adapted accordingly.