Clinicopathologic evaluation of the mitosis-specific marker phosphohistone-H3 (PHH3) in uterine smooth muscle tumors

Abstract

Uterine smooth muscle tumors (USMT) are classified into benign, uncertain malignant potential and malignant categories. Leiomyomas (LM), together with other benign histological variants, are benign. Leiomyosarcomas (LMS) are rare but are aggressive. The pathologic distinction between these entities requires histologic assessment of cytologic atypia, mitotic count and presence or absence of tumor cell necrosis (TCN). A tumor with any 2 of these 3 features is usually considered as malignant. However, even though most LMS show diffuse cytological atypia and a mitotic count of >10/10 high power fields (HPF), but up to one-third has no TCN. Determining the mitotic count has therefore become more important. This is usually performed on conventional hematoxylin and eosin (H&E) stained sections. This task is not always easy and is subjected to intra and inter-observer’s variation. Aside from tissue artifacts, the appearance of a mitotic figure may be confused with karyorrhectic nucleus or apoptotic body. This is particularly important in the case of leiomyoma with bizarre nuclei (LBN), which is a benign LM variant. The atypical nuclear appearance in this is believed to be secondary to degeneration. However, the degree of nuclear atypia in LBN is comparable to that seen in LMS, and the diagnosis from one another is heavily dependent on the mitotic count. Most LBNs have an average of <2 mitotic figures/10 HPF but degenerating, karyorrhectic nuclei makes accurate mitosis-counting difficult. Failure to distinguish between these tumors would invariably lead to diagnosis of smooth muscle tumor of uncertain malignant potential (STUMP). For the patient, it means life-long follow up. The immunohistochemical marker phosphohistone H3 (PHH3) is a mitosis-specific marker that has been studied on several cancers in which accurate mitotic count is important, but its use on USMT, especially in relation to outcome, has not been thoroughly investigated. In this study, we evaluated the performance of PHH3 on 94 USMT (48 LMS, 23 STUMP and 23 LM). The immunostain result was expressed as the highest count/10 HPF. The scores were compared with H&E mitotic counts for each category. Follow-ups were obtained. The median age of LMS patients was 52 years, while those for STUMP and LM were 42 and 45 years, respectively. The median follow-up duration of LMS, STUMP and LM were respectively 28 months (range: 1-204), 71 months (range: 5-156), and 62 months (range: 2-114). At last follow-up, 25 patients with LMS (52.1%) died of disease. Two of 23 patients (8.7%) with STUMP had a recurrence, both at 14 months after diagnosis. All with LM had a benign outcome. The median PHH3 score for LMS was 17/10 HPFs, which was significantly higher than for STUMP (3) or LM (2) (p<0.05). The corresponding H&E mitotic counts for these respective categories were 12, 3 and 1/10 HPF. To test the discriminating power of PHH3 in distinguishing benign USMT with ambiguous morphologies (LBN and STUMP) from those that were clinically malignant, two biologic groups were further delineated. Patients in group 1 all had a benign outcome (11 LBN, 9 STUMP with no recurrence) while in group 2, patients had either a recurrence or death (2 STUMP with recurrence, and 29 LMS). Median ages for group 1 and 2 patients were 44 and 53 years, respectively, and the median follow-up durations were 69 (range 5-156) and 7 (range 1-96) months, respectively. The median PHH3 score for group 1 tumors was 3/10HPF compared with 24/10 HPF for group 2. Using ROC analysis, a PHH3 score of ≥6/10 HPF is highly compatible with malignancy with sensitivity and specificity of 80.6% and 93.7%, respectively, and positive and negative predictive values of 92.5% and 83.3%, respectively. PHH3 can potentially be used to distinguish between LBN and clinically malignant USMT. Through uni- and multivariate analyses, we also determined that a PHH3 score of ≥ 29 per 10 HPF in LMS is independently associated with poor outcome (p<0.05).