Background

Pneumocystis jirovecii pneumonia (PCP) is an opportunistic infection commonly affecting immunocompromised people. Diagnosis usually requires invasive techniques to obtain respiratory specimens. Minimally invasive detection tests have been proposed, but their operating characteristics are poorly described.

Objectives

To systematically review and meta-analyse the performance of minimally invasive PCP detection tests to inform diagnostic algorithms.

Data sources

Medline, Embase, Cochrane Library (inception to 15 October 2020).

Study eligibility criteria

Studies of minimally invasive PCP detection tests were included if they contained a minimum of ten PCP cases.

Participants

Adults at risk of PCP.

Tests

Non-invasive PCP detection tests.

Reference standard

Diagnosis using the combination of clinical and radiographical features with invasive sampling.

Assessment of risk bias

Using the QUADAS-2 tool.

Methods

We used bivariate and, when necessary, univariate analysis models to estimate diagnostic test sensitivity and specificity.

Results

Fifty-two studies were included; most studies (40) comprised exclusively human immunodeficiency virus (HIV) -infected individuals; nine were mixed (HIV and non-HIV), two were non-HIV and one study did not report HIV status. Sampling sites included induced sputum, nasopharyngeal aspirate, oral wash and blood. The four testing modalities evaluated were cytological staining, fluorescent antibody, PCR and lactate dehydrogenase. Induced sputum had the most data available; this modality was both highly sensitive at 99% (95% CI 51%–100%) and specific at 96% (95% CI 88%–99%). Induced sputum cytological staining had moderate sensitivity at 50% (95% CI 39%–61%) and high specificity at 100% (95% CI 100%–100%), as did fluorescent antibody testing with sensitivity 74% (95% CI 62%–87%) and specificity 100% (95% CI 91%–100%).

Conclusion

There are several promising minimally invasive PCP diagnostic tests available, some of which may reduce the need for invasive respiratory sampling. Understanding the operating characteristics of these tests can augment current diagnostic strategies and help establish a more confident clinical diagnosis of PCP. Further studies in non-HIV infected populations are needed.

中文翻译：

---

耶氏肺孢子菌肺炎的无创诊断：系统评价和荟萃分析

背景

耶氏肺孢子菌肺炎 (PCP) 是一种机会性感染，通常影响免疫功能低下的人。诊断通常需要侵入性技术来获取呼吸道标本。已经提出了微创检测测试，但是它们的操作特性描述得很差。

目标

系统地审查和荟萃分析微创 PCP 检测测试的性能，为诊断算法提供信息。

数据源

Medline、Embase、Cochrane 图书馆（成立至 2020 年 10 月 15 日）。

研究资格标准

如果微创 PCP 检测试验包含至少 10 个 PCP 病例，则纳入研究。

参与者

有 PCP 风险的成年人。

测试

非侵入性 PCP 检测测试。

参考标准

结合临床和影像学特征与侵入性取样进行诊断。

风险偏差评估

使用 QUADAS-2 工具。

方法

我们使用双变量和必要时单变量分析模型来估计诊断测试的敏感性和特异性。

结果

纳入了 52 项研究；大多数研究 (40) 仅包括人类免疫缺陷病毒 (HIV) 感染者；九人混合（艾滋病毒和非艾滋病毒），两项非艾滋病毒，一项研究未报告艾滋病毒状况。采样部位包括诱导痰、鼻咽抽吸物、口腔冲洗液和血液。评估的四种测试方式是细胞学染色、荧光抗体、PCR 和乳酸脱氢酶。诱导痰的可用数据最多；这种方法的敏感性为 99% (95% CI 51%–100%)，特异性为 96% (95% CI 88%–99%)。诱导痰细胞学染色的敏感性为 50% (95% CI 39%–61%)，特异性为 100% (95% CI 100%–100%)，荧光抗体检测的敏感性为 74% (95% CI) 62%–87%) 和特异性 100% (95% CI 91%–100%)。

结论

有几种有前途的微创 PCP 诊断测试可用，其中一些可能会减少对侵入性呼吸采样的需求。了解这些测试的操作特征可以增强当前的诊断策略，并有助于建立更自信的 PCP 临床诊断。需要对非 HIV 感染人群进行进一步研究。