【目的】分析中国医院侵袭性真菌监测网CHIF-NET2010-2014连续五年收集的侵袭性真菌感染分离的酵母菌的流行病学及抗真菌药物敏感性，探索并完善真菌鉴定方法、药敏方法、分子分型方法，耐药机制研究，罕见真菌感染病例等相关模式研究，从而初步构建一套关于临床重要真菌相关的科学研究方法体系。

【方法】CHIF-NET 2010-2014五年参加单位共65家，所有收集的菌株在中心实验室统一使用分子生物学方法鉴定到种，使用纸片扩散法测定对氟康唑、伏立康唑的药物敏感性；对Vitek MALDI-TOF MS 和Bruker Biotyper MALDI-TOF MS两套质谱系统进行评价研究；以罗伦隐球菌为例评价传统酵母菌鉴定方法Vitek 2 Compact真菌鉴定卡和API 20C AUX试剂条的鉴定准确性；以微量肉汤稀释法为金标准评价商品化真菌药敏试验试剂盒ATB® FUNGUS 3的性能；以隐球菌为例研究ITS分型、MLST等分子分型技术的鉴别能力；以季也蒙念珠菌为例，研究ERG11基因与唑类耐药的关系；以橡树念珠菌为例，对罕见真菌感染病例的临床特点进行首次报道。

【结果】CHIF-NET 2010-2014收集的9673株菌株共包含70个菌种，与分子鉴定结果相比，原始鉴定结果的准确率为85.9%。在所有菌种中，念珠菌属最为常见（构成比占91.3%），其次为隐球菌属（6.8%）、毛孢子菌属（0.8%）和其他少见菌属（1.2%）。在念珠菌属中，白念珠菌仍是最常见的菌种（44.9%），非白念珠菌的比例超过了50%。Vitek MS v2.0能将96.7%的菌株正确鉴定至种水平；而Bruker Biotyper MALDI-TOF MS鉴定酵母菌的准确率可达98.8%，特别是念珠菌属的鉴定能力更优（准确率达99.4%）。根据纸片法药敏结果，念珠菌属对氟康唑和伏立康唑总体耐药率为8.8%和3.8%，白念珠菌最为敏感，耐药率不到1%；近平滑念珠菌复合体敏感性也相对较高，达90.8%和97.5%；光滑念珠菌对氟康唑和伏立康唑的耐药率（或非野生型）分别为19.0%和18.7%，而热带念珠菌对两种药物的耐药率五年间上升较为明显，从5%左右均上升到20%以上。Vitek 2 Compact系统鉴定26株菌为罗伦隐球菌，通过ITS测序仅确认两株（7.7％）为罗伦隐球菌，API 20C系统表现出相对更好的鉴定准确度（83.3％）。对于氟康唑和伏立康唑，ATB® FUNGUS 3肉眼判读和微量肉汤稀释法的总分类一致率（CA）分别是95.5%和95.4%，相比之下，ATB全自动仪器判读的准确性明显下降，对氟康唑和伏立康唑的总CA分别为77.2%和73.3%。MLST分型可将303株菌新型隐球菌分为12个ST型别，89.8%为ST5型，ST31型是第二大型别；7株格特隐球菌基因变异较大，分为5个ST型别。对164株季也蒙念珠菌的ERG11基因进行扩增和测序，通过比较氨基酸序列型的多态性，共鉴定出16个基因型，其中一些型别与唑类药物敏感性密切相关。对氟康唑耐药的橡树念珠菌（C. quercitrusa）作为人类致病菌进行了首次报道，并确认这3例菌株是来自H1医院ICU病房的院内感染暴发。

【结论】CHIF-NET2010-2014五年连续监测积累了大量珍贵的菌株及临床信息资源。本项研究是国内真菌领域内时间跨度最大、参与医院最多、涉及菌株数量最多的流行病学和药物敏感性研究，填补了国内相关领域的空白，为中国侵袭性真菌感染诊治提供重要参考。初步搭建了国内关于真菌鉴定方法、药敏方法、分子分型方法、耐药机制研究、罕见感染病例报道等临床重要真菌相关的科学研究方法体系，但该体系的成熟与完善仍任重而道远。

【Objective】To investigate the epidemiology and antifungal susceptibility of yeast isolates causing invasive fungal infections collected from China Hospital Invasive Fungal Surveillance Net (CHIF-NET) study 2010-2014, and to establish a whole scientific research system of clinical important fungi including identification, antifungal susceptibility, typing, resistant mechanism and rare infection cases.

【Methods】There were 65 hospitals participated in the CHIF-NET surveillance 2010-2014. All 9673 isolates collected were identified to species level by unique molecular identification methods, and antifungal susceptibility testing to fluconazole and voriconazole was performed by disk diffusion method. The evaluation of Vitek MALDI-TOF MS and Bruker Biotyper MALDI-TOF MS were performed in the isolates collected from CHIF-NET 2011 and 2013, respectively.Vitek 2 Compact YST card and API 20C AUX were used to compare the accuracy in identifyingCryptococcus laurentiiisolates. The capability of the popular used commercial kit ATB® FUNGUS 3 in antifungalsusceptibility testing was evaluated according to the golden method both microdilution. Several typing methods such as ITS and MLST were compared using Cryptococcusstrains. The relation between ERG11 genes and azoles resistance were investigated using Candida guilliermondii isolates.To firstly report the rare fungal infection case causing by C.quercitrusa.

【Results】Of all 9673 isolates collected from CHIF-NET 2010 to 2014, 70 species were identified totally. Comparing to molecular identification results, only 85.9% of initial identifications were correct. Candida species was predominant (91.3%), followed by Cryptococcus species (6.8%), Trichosporon species (0.8%) and other yeast species (1.2%). Candida albicans was the commonest Candida species (44.9%), whilst the non-albicans took more than 50%. Vitek MALDI-TOF MS v2.0identified96.7% isolates correctly to the species level, but Bruker Biotyper MALDI-TOF MS showed a better result of 98.8%, especially the Candida spp.(99.4%). By using CLSI disk diffusion testing,the resistant rates of Candida spp.tofluconazole and voriconazole were 8.8% and 3.8%, respectively.Candida albicans was the most sensitive species which resistant rate was less than 1%;C. parapsilosis complex also showed relative higher sensitivity of 90.8% tofluconazole and97.5%to voriconazole; the resistant (non-WT) rates of C. glabratatofluconazole and voriconazole were 19.0% and 18.7%, respectively; the resistance to both two antifungals increaseddramatically during the past five years, from 5% to above 20%.There were 26 Cryptococcus laurentiiisolates were identified by Vitek 2 Compact systems, however only two isolates (7.7%) were confirmed by ITS sequencing; whileAPI 20C AUX showed a better accuracy of 83.3%. The CA of ATB® FUNGUS 3 visual reading tofluconazole and voriconazole were95.5% and 95.4%, respectively; while the CA of ATB® FUNGUS 3 automatic reading were 77.2% and 73.3%, respectively. There were 12 ST types of Cryptococcus neoformans complexusing MLST method, ST5 accounted for 89.8%, followed by ST31; there were 5 ST types in Cryptococcus gattii. TheERG11 genes were amplified and sequenced in 164 Candida guilliermondiiisolates, 16 genotypes were identified, some of them were closelyrelevant with azoles resistance. The C. quercitrusaisolates which caused human infections were firstly reported inthis study, and nosocomial infection outbreak of these three C. quercitrusastrains were confirmed in the ICU ward of Ha’erbin hospital.

【Conclusions】 The five years’ CHIF-NET study has accumulated valuable micro-organism and clinical information sources. This study is a very important epidemiology and antifungal susceptibility study which involved the largest numbers of hospitals, isolates and longest time, It provided useful data to the field of invasive fungal infections previous we lack of knowledge. This study built up a scientific research system of clinical important fungi including identification, antifungal susceptibility, typing, resistant mechanism and rare infection cases, but there is still a long way to go.