分析肉芽肿性小叶性乳腺炎(GLM)患者的病原菌分布，探讨克氏棒状杆菌(克氏菌)感染与GLM患者外周血血象指标变化的关系。

回顾性分析2017年10月至2021年6月武汉大学人民医院乳腺甲状腺中心诊治的81例GLM患者临床资料和纳米孔测序获得的病原菌检测结果。根据病原菌感染分为3组：克氏菌组(50例)、非克氏菌组(19例)和阴性组(12例)，并将50例克氏菌组患者按病程阶段分为早期(21例)和晚期(29例)。计数资料以绝对数表示。采用Kruskal-Wallis H检验或Mann-Whitney U检验比较3组患者外周血白细胞计数和中性粒细胞与淋巴细胞比值(NLR)的差异，受试者工作特征曲线检验外周血白细胞计数和NLR对克氏菌感染的鉴别诊断价值，Kappa检验用于外周血白细胞计数和NLR与纳米孔测序的一致性检验。

(1)85.2%(69/81)患者的样本中检测到病原菌，其中61.7%(50/81)患者存在克氏菌感染。(2)克氏菌组、非克氏菌组和阴性组患者的外周血白细胞计数异常率分别为48.0%(24/50)、3/19和4/12，组间比较差异有统计学意义(χ²＝6.192，P＝0.045)，克氏菌组高于非克氏菌组(χ²＝5.997，Bonferroni校正后P＝0.043)。3组患者的外周血白细胞计数分别为9.09(7.50，12.02)×10⁹个/L、7.58(6.47，8.20)×10⁹个/L和7.62(6.99，9.84)×10⁹个/L，组间比较差异有统计学意义(H＝8.748，P＝0.013)，克氏菌组高于非克氏菌组(Z＝－2.861，Bonferroni校正后P＝0.012)。早期患者的外周血白细胞异常率和白细胞计数均低于晚期患者[28.6%(6/21)比62.1%(18/29)，χ²＝5.476，P＝0.019；7.50(6.74，10.65)×10⁹个/L比9.69(8.76，13.67)×10⁹个/L，Z＝3.155，P＝0.002]。(3)3组患者的NLR分别为3.47(2.67，5.73)、2.95(2.04，3.35)和3.06(1.87，3.37)，组间比较差异有统计学意义(H＝9.417，P＝0.009)，克氏菌组NLR高于非克氏菌组(Z＝－2.556，Bonferroni校正后P＝0.039)。早期患者的NLR低于晚期患者[3.27(1.89，3.89)比4.45(3.13，6.86)，Z＝2.251，P＝0.024]。(4)外周血白细胞计数的ROC曲线下面积为0.724(95%CI：0.599~0.849，P＝0.004)，Kappa系数为0.359，当外周血白细胞计数>8.33×10⁹个/L时诊断克氏菌感染的敏感度、特异度、准确率、阳性预测值、阴性预测值、阳性似然比、阴性似然比、约登指数和假阴性率分别为54.0%(27/54)、18/19、55.6%(45/81)、96.4%(27/28)、43.9%(18/41)、10.260、0.507、0.487、46.0%(23/50)。NLR的ROC曲线下面积为0.701(95%CI：0.577~0.826，P＝0.010)，Kappa系数为0.313，当NLR>3.13时诊断克氏菌感染的敏感度、特异度、准确率、阳性预测值、阴性预测值、阳性似然比、阴性似然比、约登指数和假阴性率分别为68.0%(34/50)、13/19、58.0%(47/81)、85.0%(34/40)、44.8%(13/29)、2.153、0.467、0.364、32.0%(16/50)。

克氏菌与GLM关系密切，伴有克氏菌感染的GLM患者外周血炎症指标更高。临床医师在GLM的诊疗过程中应该重视病原菌的检测，为患者提供更加个体化、精准化的治疗方案。

To investigate the prevalence of pathogenic bacteria in patients with granulomatous lobular mastitis (GLM) and explore the relationship between Corynebacterium kroppenstedtii (CK) infection and peripheral hematological parameters.

The clinical data and pathogenic bacteria identification by nanopore sequencing of 81 patients with GLM in the Renmin Hospital of Wuhan University from October 2017 to June 2021 were reviewed retrospectively. The enrolled patients were divided into 3 groups according to the pathogenic results: CK-positive group (n=50), CK-negative group (n=19) and non-pathogenic group (n=12). The CK-positive group was further divided into the early-stage subgroup (n=21) and the advanced-stage subgroup (n=29). Counting data is expressed in absolute numbers. The Kruskal-Wallis H test or Mann-Whitney U test was used to compare peripheral white blood cell (WBC) count and neutrophil-to-lymphocyte ratio (NLR) among three groups. The receiver operating characteristic (ROC) curve were plotted to analyze the value of WBC count and NLR in detecting CK infection, and the Kappa test was used to analyze the consistency of WBC count and NLR with nanopore sequencing.

Approximately 85.2% (69/81) of the patients were tested positive by nanopore sequencing, of which 61.7% (50/81) were CK-positive. The proportion of patients with abnormal WBC count in CK-positive group, CK-negative group and non-pathogenic group was 48.0% (24/50), 3/19 and 4/12, respectively, indicating a significant difference (χ²=6.192, P=0.045). The proportion of patients with abnormal WBC count in CK-positive group was significantly higher than that in CK-negative group (χ²=5.997, P=0.043 after Bonferroni correction). The peripheral WBC count in CK-positive group, CK-negative group and non-pathogenic group was 9.09 (7.50, 12.02)×10⁹ cells/L, 7.58 (6.47, 8.20)×10⁹ cells/L, and 7.62 (6.99, 9.84)×10⁹ cells/L, respectively, indicating a significant difference (H=8.748, P=0.013). The peripheral WBC count in CK-positive group was significantly higher than that in CK-negative group (Z=-2.861, P=0.012 after Bonferroni correction). The proportion of patients with WBC abnormality and WBC count in early-stage subgroup were significantly lower than those in advanced-stage subgroup [28.6% (6/21) vs 62.1% (18/29), χ²=5.476, P=0.019; 7.50 (6.74, 10.65)× 10⁹/L vs 9.69 (8.76, 13.67)×10⁹/L, Z=3.155, P=0.002]. (3) The NLR in CK-positive group, CK-negative group and non-pathogenic group was 3.47 (2.67, 5.73), 2.95 (2.04, 3.35) and 3.06 (1.87, 3.37), respectively, suggesting a significant difference (H=9.417, P=0.009). The NLR in CK-positive group was significantly higher than that in CK-negative group (Z=-2.556, P=0.039 after Bonferroni correction). The NLR in early-stage subgroup were significantly lower than that in advanced-stage subgroup [3.27 (1.89, 3.89) vs 4.45 (3.13, 6.86), Z=2.251, P=0.024]. (4) The area under the ROC curve of WBC count was 0.724 (95%CI: 0.599-0.849, P=0.004), and the Kappa coefficient was 0.359. If WBC count > 8.33×10⁹ cells/L, its sensitivity, specificity, accuracy, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Youden index and false negative rate in diagnosing CK-infection were 54.0% (27/54), 94.7% (18/54), 55.6% (45/81), 96.4% (27/28), 43.9% (18/41), 10.260, 0.507, 0.487 and 46.0% (23/50), respectively. The area under the ROC curve of NLR was 0.701 (95%CI: 0.577-0.826, P=0.010), and the Kappa coefficient was 0.313. If NLR>3.13, the sensitivity, specificity, accuracy, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Youden index and false negative rate in diagnosing CK-infection were 68.0% (34/50), 68.4% (13/19), 58.0% (47/81), 85.0% ( 34/40), 44.8% (13/29), 2.153, 0.467, 0.364 and 32.0% (16/50), respectively.

CK is closely related to GLM, and GLM patients with CK infection show high level in hematological inflammation parameters. Clinicians should pay attention to pathogen detection in the diagnosis and treatment of GLM and provide individualized precise regimens.