垂体促甲状腺激素腺瘤中SSTR表达谱及全数字切片自动图像分析的探索性研究

  目的  探索垂体促甲状腺激素（thyroid stimulating hormone，TSH）腺瘤中生长抑素受体（somatostatin receptor，SSTR）的表达以及全数字切片自动图像分析技术的应用。  方法  分析2013年1月至2019年12月北京协和医院的37例垂体TSH腺瘤的石蜡包埋组织标本及临床病理资料，利用各型SSTR1-SSTR5抗体进行免疫组织化学染色，分别进行人工判读和全数字切片自动图像分析，比较各型SSTR表达与性别、年龄、肿瘤大小、Ki-67指数以及侵袭的相关性，并对比人工判读和软件自动分析的一致性。   结果  垂体TSH腺瘤中SSTR1、SSTR2、SSTR4和SSTR5多为2分和3分表达，其中3分占比均超过50.0%（4型SSTR占比依次分别为59.5%、78.4%、64.9%和56.8%）。SSTR1表达与侵袭呈明显负相关（人工r=-0.417，P=0.010；软件r=-0.406，P=0.013），并与SSTR2表达具有显著相关性（人工r=0.420，P=0.010；软件r=0.338，P=0.040）。各型SSTR表达的人工判读与图像自动分析结果均显示了较好的一致性。  结论  垂体TSH腺瘤肿瘤组织中SSTR1、SSTR2、SSTR4和SSTR5均有较高水平的表达，利用基于数字化全切片的图像自动分析技术可获得与病理医师人工判读相近的评估效果。 

Somatostatin receptor profile in pituitary thyroid stimulating hormone-producing adenoma and an exploratory study of automated image analysis for the digitized whole slide

  Objective  To evaluate the somatostatin receptor (SSTR) profile in pituitary thyroid stimulating hormone (TSH)-producing adenoma and to perform an exploratory study of automated image analysis for the digitized whole slide.   Methods  This retrospective study of pituitary TSH-producing adenoma including cases admitted in Peking Union Medical College Hospital from January 2013 to December 2019 analyzed the SSTR profile in the archived formalin-fixed, paraffin-embedded tissues using immunohistochemistry (IHC). Scoring of IHC staining was performed using both conventional microscopic evaluation and automated image analysis for the digitized whole slide. We compared the correlation coefficient between the two methods and analyzed the correlation between SSTR expression and clinical-pathological characteristics (including sex, age, tumor size, Ki-67 index, and invasiveness).   Results  Thirty-seven cases of TSH-producing adenomas were included; most had a high expression (Score 3 or 2) of SSTRs except SSTR3. Overall, more than 50% cases scored 3 (59.50%, 78.40%, 64.9%, and 56.80% of the cases scored 3 for SSTR1, SSTR2, SSTR4, and SSTR5, respectively). SSTR1 expression showed a significantly negative correlation with invasiveness (r=-0.417, P=0.010 for conventional scoring; r=-0.406, P=0.013 for automated image analytic scoring), whereas SSTR2 expression showed significant correlation (r=-0.420, P=0.010 for conventional scoring; r=-0.338, P=0.040 for automated image analytic scoring). All SSTRs showed good consistence between two scoring methods.  Conclusions  In this TSH-producing adenoma case series, we identified a high expression of SSTR1, SSTR2, SSTR4, and SSTR5 and applied automated image analysis for the digitized whole slide of TSH-producing adenoma; an effect similar to that of conventional scoring was acquired.