The CAP: Your resource for artificial intelligence information.

Artificial intelligence (AI) has become increasingly studied and implemented in some pathology departments; but how might AI models help change medicine? While technology leaders are calling for a pause on developing AI systems until there is greater confidence "that their effects will be positive and their risks will be manageable," (Pause Giant AI Experiments: An Open Letter. March 22, 2023. https://futureoflife.org/open-letter/pause-giant-ai-experiments/) could breakthrough innovations in AI make medicine and physicians better? (There's One Hard Question My Fellow Doctors and I Will Need to Answer Soon. July 6, 2023. https://www.nytimes.com/2023/07/06/opinion/artificial-intelligence-medicine-healthcare.html) This discussion will provide an overview of AI, its current role in pathology practice, what pathologists need to know, and what the College of American Pathologists is doing for pathologists in this area.

Histopathologic Patterns of Colitis in Patients With Impaired Renal Function.

OBJECTIVES: To characterize the histopathologic features of colitis in patients with impaired renal function. METHODS: We retrospectively identified 413 patients who underwent colonoscopic evaluation for colitis between 2011 and 2015. Patients were divided into four groups based on estimated glomerular filtrate rates. Patients with impaired renal function were compared to overall and age-matched patients with normal renal function. RESULTS: Compared to a preponderance of inflammatory bowel disease (33%) and lymphocytic colitis (9.6%) in patients with normal renal function, ischemic colitis (58%) was the predominant histopathologic pattern in the patients with impaired renal function. Infectious colitis was the second most common pattern (20.8%), with Clostridium difficile and cytomegalovirus infections being more frequent. Medication-induced injury was the third most common pattern, with crystal-associated injury being the exclusive pattern found in this study. CONCLUSIONS: Colitis in patients with impaired renal function is etiologically distinct from that seen in patients with normal renal function.

Pitfalls in salivary gland fine-needle aspiration cytology: lessons from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology.

CONTEXT: We use data from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology to identify common diagnostic errors in salivary gland fine-needle aspiration (FNA). OBJECTIVE: To identify salivary gland FNA cases with poor performance characteristics in the Nongynecologic Cytology Program surveys, so that the most common diagnostic pitfalls can be avoided. DESIGN: A retrospective review of the College of American Pathologists Nongynecologic Cytology Program's cumulative data from 1999 to 2003 revealed the most common false-positive and false-negative interpretations on FNA for common salivary gland lesions. Slides that performed poorly were then reviewed to identify the cytologic characteristics that may have contributed to their poor performance. RESULTS: A total of 6249 participant responses with general interpretations of benign (n = 4642) or malignant (n= 1607) were reviewed. The sensitivity and specificity of the participant responses for correctly interpreting the cases as benign or malignant were 73% and 91%, respectively. Benign cases with the highest false-positive rates were monomorphic adenoma (53% false-positive), intraparotid lymph node (36%), oncocytoma (18%), and granulomatous sialadenitis (10%). Malignant cases with the highest false-negative rates were lymphoma (57%), acinic cell carcinoma (49%), low-grade mucoepidermoid carcinoma (43%), and adenoid cystic carcinoma (33%). Selected review of the most discordant individual cases revealed possible explanations for some of the interpretative errors. CONCLUSIONS: These data confirm the difficulty associated with interpretation of salivary gland FNA specimens. Cytologists should be aware of the potential false-positive and false-negative interpretations that can occur in FNAs from this organ site in order to minimize the possibility of diagnostic errors.

Fluids--good and bad actors: observations from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology.

CONTEXT: Body cavity fluid examination presents a common and sometimes difficult diagnostic challenge in daily cytology practice. Separating benign from malignant cellular changes may require meticulous screening, careful scrutiny of cellular features, and an understanding of the range of reactive changes. We use the data from the College of American Pathologists (CAP) Interlaboratory Comparison Program in Nongynecologic Cytology (NGC) to identify characteristics of fluids that place them at opposite ends of the diagnostic spectrum. OBJECTIVE: To assess the features of individual body cavity fluid slides that demonstrated good performance characteristics and compare them to slides that were poor performers. DESIGN: A databank of 10 396 laboratory responses, including a variety of malignant and benign cases obtained from 1997 through 2001, was used to select cases. A cumulative slide history was used to identify slides that performed well or poorly in each reference diagnosis. Cases were confirmed by consensus of 4 CAP Cytopathology Resource Committee members. Observations and characterizations of good and bad performers in each category were recorded and summarized. RESULTS: Percentage of concordance of poor performers ranged from 0% to 58%. Conversely, good performers were identified with high concordance of laboratory diagnosis in each reference category (>80%). Several patterns emerged. Poorly performing cases of adenocarcinoma consisted of slides with rare tumor cells, hypercellular malignant cases without 2 cell populations, and cases with single cells. Poor performance in confirmed squamous cell carcinoma cases related to rare cells without keratinization. Small cell carcinoma and melanoma cases performed poorly when there were few malignant cells. Lymphoma cases demonstrated poor performance when there were abundant pleomorphic lymphoid cells or when rare Reed- Sternberg-like cells were present. Reactive or negative slides performed best with a polymorphous population; poor performers were those with a predominant lymphocyte population mistaken for a hematopoietic neoplasm. CONCLUSION: Close attention to classic cytologic criteria and careful examination of slides may enhance the educational experience of participants and the performance characteristics of body cavity fluid specimens in the CAP NGC program. Lessons from bad actors in the CAP NGC program may increase awareness of potential diagnostic problems in daily practice or help identify areas for laboratory quality improvement.

Detection of adenocarcinoma in situ of the cervix in Papanicolaou tests: comparison of diagnostic accuracy with other high-grade lesions.

CONTEXT: Adenocarcinoma in situ of the cervix is a recently recognized interpretation in the Bethesda 2001 system. Although specific morphologic criteria have been published, recognizing this entity is still difficult. OBJECTIVE: To compare pathologists' ability to correctly identify and categorize adenocarcinoma in situ with their ability to identify and categorize adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma. DESIGN: Pathologists' reviews in the 2001 and 2002 College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology Program, an interlaboratory comparison program for gynecologic cytology, were examined. Cases were usually reviewed by multiple pathologists. False-negative rates, the percentage of reviews with exact agreement with reference interpretations, and the percentage of cases in which all reviews were in exact agreement with the reference interpretation for adenocarcinoma in situ, adenocarcinoma, high-grade squamous intraepithelial lesion, and squamous cell carcinoma were compared. RESULTS: A total of 213 reviews of cases categorized as adenocarcinoma in situ were compared with 2821 reviews of adenocarcinoma, 7535 reviews of high-grade squamous intraepithelial lesion, and 1886 reviews of squamous cell carcinoma. The false-negative rate for adenocarcinoma in situ (11.7%) was significantly higher than that for high-grade squamous intraepithelial lesion (4.6%, P <.001) and squamous cell carcinoma (3.3%, P <.001) but not for adenocarcinoma (8.9%, P =.16). Of all the reviews of adenocarcinoma in situ cases, 46.5% were interpreted specifically as adenocarcinoma in situ, compared to 72.2% of reviews of adenocarcinoma, 73.2% of high-grade squamous intraepithelial lesion, and 75.1% of squamous cell carcinoma. No individual case of adenocarcinoma in situ was always specifically recognized as adenocarcinoma in situ; 26.5% of cases of adenocarcinoma were specifically recognized as such in all reviews. Findings were similar with and without the inclusion of high-grade squamous intraepithelial lesion/carcinoma, not otherwise specified, as an acceptable review interpretation for cases of adenocarcinoma, squamous cell carcinoma, and high-grade squamous intraepithelial lesion. CONCLUSION: These data from expert-referenced and biopsy-proven cases suggest that adenocarcinoma in situ is not as easily recognized or categorized as other serious diagnoses.