Cart before the horse: use of Aspergillus PCR to increase the diagnostic yield from BAL in hematological patients at risk of invasive aspergillosis

Lower respiratory tract infections are a major cause of morbidity and mortality in hematological patients [1–3]. Accurate diagnosis is essential because there are a broad range of infective and noninfective etiologies. Computed tomography (CT) evaluation is an essential component of early evaluation as radiographic features may be suggestive of etiology [4]. However, while CT findings may be suggestive, they are not diagnostic and invasive sampling is required to confirm a diagnosis [5,6].

Marchesi et al. report a 11% (85/769) incidence of pulmonary infiltrates requiring hospitalization in patients treated for hematological malignancy, excluding allogeneic transplant, which is similar to others [3,7]. 47 of these patients underwent bronchoscopy with lavage and specimens were then subjected to a prespecified diagnostic panel including those for the investigation of Invasive Aspergillosis (IA); fungal culture, an Aspergillus PCR, and galactomannan EIA (GM) in an attempt to accurately diagnose the disease etiology. Sampling of the lower respiratory tract via bronchoalveolar lavage (BAL) significantly improves isolation of an infectious agent in immunosuppressed patients with pulmonary infiltrates compared to sampling of the upper respiratory tract alone [8,9]. But, despite improving diagnostic gain, positive yields from BAL still remain less than 60% [9,10]. Similarly, in this cohort, described by Marchesi et al., in only 65% of patients, who underwent bronchoscopy, a causative organism was identified [7].

It is important to acknowledge that in this cohort, the time from diagnosis of infiltrate to bronchoscopy was a median of 3 days (range 1–13) [7]. Early access to bronchoscopy is critical. The positive yields from BAL in HSCT recipients are reported between 42 and 66% and are greatest when performed within the first 24 h of presentation [8]. In addition to improved yield, timely bronchoscopy also provides information to refine empiric antimicrobial therapy, an important component of antimicrobial stewardship. Clinical pathways that support early access to bronchoscopy are important in hematology units.

While the spectrum of infectious agents in hematology patients with pulmonary infiltrates is broad, invasive pulmonary aspergillosis (IA) is the most commonly encountered fungal infection and is associated with high morbidity and mortality [11–15]. While it is known that timely diagnosis of IA leads to earlier initiation of appropriate and targeted antifungal therapy and improved clinical outcomes, the accurate diagnosis of IA has posed major challenges [11,13,14,16]. Marchesi et al., like others, have set out to provide evidence that incorporation of an Aspergillus PCR into IA diagnostic algorithms will lead to increased diagnosis, earlier initiation of therapy and improved outcomes.

In current definitions of IFD developed by the EORTC/MSG, a ‘proven’ diagnosis of IA requires either a histological confirmation of fungal invasion into tissue or positive culture of diseased tissue [15]. This typically requires invasive sampling which is often difficult and even contraindicated in critically unwell patients. Furthermore, cultures take days to cultivate and yields rarely exceed 50% [12,17]. Galactomannan (GM) has been incorporated into the EORTC/MSG definitions for over 15 years and multiple studies have demonstrated good sensitivity for the detection of IA, particularly when performed on BAL specimens in patients with hematological malignancy [12,18]. However, sensitivity of the result may be reduced in non-neutropenic patients and in the presence of antifungal therapy, including antifungal prophylaxis, and debate remains regarding the optimal optical density threshold that should define a positive result [19,20]. Furthermore, false positive results have been reported in the presence of cross-reacting fungal pathogens including Pencilillium and antibiotics including pipericillin-tazobactam. Despite these limitations, the estimated sensitivity of GM in BAL in HM patients is between 80–90% with a specificity of 89–94% [18,21]. In the Marchesi et al. cohort, both serum and BAL GM were used for the diagnosis of IA [7]. Serum GM does not perform as well in the presence of mold active antifungal prophylaxis or treatment and is best used as part of a diagnostic workup in such patients rather than as routine surveillance [21–23]. The findings of Marchesi et al. support this approach as only 2 of the 19 cases of probable IA had positive serum GM (defined as two consecutive samples with an OD ≥0.5 or a single sample ≥0.7) [7].

Aspergillus PCR has been a promising diagnostic tool available for over 20 years but due to concerns regarding lack of standardization and validation has been repeatedly omitted from the EORTC/MSG and other diagnostic algorithms. Since 2006, international collaboration has resulted in a global performance standard between laboratories which has subsequently paved the way for the development of commercial assays. As it stands, Aspergillus PCR is the only indirect mycological test that has independently available control material and international QC schemes to calibrate and allow impartial comparison of different assays [24]. In a systematic review by Avni et al. evaluating the diagnostic accuracy of PCR in BAL compared to GM, PCR demonstrated a statistically higher sensitivity with no loss of specificity when a GM optical density threshold of ≥1.0 was considered positive [22]. This study by Marchesi et al. found that the addition of Aspergillus PCR testing to the evaluation of the BAL fluid in hematology patients with pulmonary infiltrates increased the diagnostic yield by over 50% with an additional seven cases meeting a definition of probable IA as defined by the investigators [7]. All of these patients had negative BAL fungal culture, serum, and BAL GM but were PCR positive. The argument presented by these investigators, and others, is that with a significant body of evidence to support the standardization and validation of PCR, this test can confidently be incorporated into IA diagnostic algorithms. In the cases presented by Marchesi et al., targeted antifungal therapy was administered to this subset of patients based on the results of CT scan and a positive PCR with a satisfactory clinical outcome reported at 90days supporting this hypothesis. In a review of 166 patients with hematological malignancy and pulmonary infiltrates, Heng et al. identified eight (7%) additional patients with a positive PCR but negative GM who would be upgraded from a possible to probable diagnosis were an update of the EORT/MSG definitions to incorporate PCR [25]. It is important to note again that the optical density chosen in the Marchesi et al. analysis was a cut off of >1 in BAL fluid. There is considerable discussion regarding the optimal threshold to consider a GM in BAL positive with demonstration of inverse relationship between sensitivity and specificity such that a higher threshold increases specificity at the expense of sensitivity [12,25]. It may be that different thresholds are applied to different clinical scenarios depending on the presence of antifungal use, neutropenia, and degree of immunosuppression. Although the range of optical density values are not reported by Marchesi et al., it is likely that had they employed a lower positive threshold of GM (i.e. between 0.5 and 0.8 as others have) there would have been a greater positive concordance between GM and PCR.

One of the limitations of the Aspergillus PCR is its poor discriminatory power between detecting colonization of the respiratory tract versus invasive disease [12,25]. But, while GM cross reacts with other fungal species detecting a broad range of fungi, PCR is the only nonculture-based assay that can be used to identify Aspergillus to a species level [26]. Understanding the limitations and strengths of both the GM and the PCR highlight the value of performing these tests together with a combination, demonstrating improved sensitivity without loss of specificity [22]. Perhaps the strongest argument for utilizing these tests concurrently is the calculated negative predictive value when both the tests are negative of 99% [22] allowing clinicians to effectively rule out IA.

While we await the outcomes of IA diagnostic algorithm revision, PCR technology has now expanded to include a multiplex real-time PCR that detects genes associated with azole resistance, which can be performed directly on BAL specimens accurately distinguishing between wild-type and azole-resistant Aspergillus species [27,28]. With azole resistance emerging in A. fumigatus, both in patients with and without prior exposure to azole therapy, early identification of resistant isolates is of increasing importance [29]. Traditional culture and susceptibility methods are time-consuming and early identification with rapid diagnostics is likely to lead to improved patient outcomes. Where these new technologies, including the AsperGenius assay, will fit into algorithms is yet to be determined but there is little doubt that clinicians faced with patients with possible invasive fungal disease developing during azole prophylaxis or in areas where azole-resistant environmental aspergillus isolates are prevalent will appreciate the rapid access to reliable azole susceptibility testing, which can be performed directly on BAL material.

The diagnosis of IA in immunocompromised patients remains a challenge. Marchesi et al. have described a cohort of hematology patients that, similar to others, have a significant prevalence of lower respiratory tract disease, a significant proportion of which is made up of IA. Molecular diagnostics for Aspergillus have significant potential to improve clinical outcomes by increasing the test turn around time, improving the sensitivity of BAL fluid analysis when compared to the gold standard of culture and microscopy and increasing the specificity of other indirect mycological tests. We await the updated EORT/MSG definitions of IFD to see whether improved standardization and validation and understanding of the factors impacting the test’s performance have been sufficient to secure its inclusion as a diagnostic criterion. Molecular diagnostics for IA are rapidly advancing and implementation by laboratories serving hematology centers will be a key factor in improving patient outcomes.

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article online at https://doi.org/10.1080/10428194.2017.1330479.