Overdiagnosis of HSIL on cervical biopsy: errors in p16 immunohistochemistry implementation

Summary In 2012, the lower anogenital squamous terminology (LAST) project introduced new nomenclature for human papillomavirus–related squamous lesions of the lower genital tract: low-grade and high-grade squamous intraepithelial lesion (LSIL and HSIL). Biomarker p16 INK4a immunohistochemistry (IHC) was also recommended to assist classification: block-like positive staining supports the diagnosis of HSIL. We aim to assess the impact of LAST recommendations on our practice as well as to identify challenges and errors in p16 IHC implementation. We studied 262 cervical biopsies meeting 3 criteria: (1) HSIL diagnosis; (2) p16 IHC performed at time of diagnosis; and (3) patient's follow-up more than 12 months, including cervical cytology, biopsy, and excision. Among these patients, subsequent loop electrosurgical excision procedure and surveillance revealed 163 HSILs (62%), 28 LSILs (11%), and 71 "nondysplastic changes" (27%). We reviewed the latter 2 groups' original hematoxylin and eosin and p16 IHC slides. The diagnosis of HSIL was confirmed in 49 cases (49%), whereas 50 (51%) were reclassified as LSIL (n=46) or negative for dysplasia (n=4). These cases were initially overdiagnosed as HSIL because pathologists (1) overused p16 IHC on unequivocal LSIL (n=27) or (2) upgraded questionable lesions to HSIL based on nonblock p16 staining patterns (patchy or focal, n=23). To implement LAST recommendations successfully, we advocate judicious use of p16 in the designated circumstances and careful interpretation of staining patterns in the context of morphology. A standardized threshold for p16 positivity and supplementary guidance will help clarify the biomarker's utility and will facilitate LAST implementation in routine practice.