COUNTERPOINT
Consider the following clinical scenario. A 27-year-old female patient with a history of Crohn’s disease who had recently been treated with a prolonged course of high-dose corticosteroids is admitted to the hospital with acute onset of fevers and severe abdominal pain. A computed tomography (CT) scan of her abdomen and pelvis reveals a perforation in her ascending colon with associated intra-abdominal fluid collection. She is started on broad-spectrum antibiotic therapy with intravenous cefepime and metronidazole and is taken to the operating room for abdominal washout and colonic resection with primary anastomosis. The patient’s fever persists postoperatively, and on hospital day 5, she develops recurrent emesis and acute-onset hypoxia. A CT of her chest, abdomen, and pelvis reveals bilateral ground-glass pulmonary opacities, findings consistent with a small bowel obstruction and multiple intra-abdominal fluid collections. A serum β-d-glucan (BDG) assay is performed as part of the workup for the patient’s fevers, ground-glass opacities, and hypoxia and returns with a positive result at 120 pg/ml.
The case noted above—which describes a patient with a number of potential causes for an elevated serum BDG level, including recent serosal exposure to surgical gauze, the likely presence of Candida spp. within her intra-abdominal fluid collections, and potential Pneumocystis jirovecii pneumonia (PJP) due to prolonged high-dose corticosteroid use—highlights the primary limitation of the serum BDG assay as a diagnostic test for PJP: its lack of specificity. As we will demonstrate in this counterpoint, the test too often yields a false-positive result. This is especially common in patients with a low pretest probability for PJP, a scenario commonly encountered in many institutions (1). We will also illustrate some specific clinical settings where use of this test may be helpful when considering a diagnosis of PJP.
Serum enzyme immunoassay kits that assess for the presence of (1,3) β-d-glucan, a polymer of glucose found in the outer cell wall of many pathogenic fungi (2), have been lauded as an ideal method for diagnosing PJP as they do not require patient participation or invasive procedures for sample collection. Although elevated levels of serum BDG strongly correlate with the presence of PJP in both HIV-infected and non-HIV-infected patients (3, 4), the currently available BDG assays lack the specificity needed to establish a diagnosis of PJP with this test alone.
Three meta-analyses have now evaluated the clinical performance of the serum BDG assay in the diagnosis of PJP (5–7). These studies demonstrated that serum BDG assays reliably detected the presence of PJP, with a range of pooled sensitivities of 91 to 96% between the three meta-analyses. These studies also revealed the greatest weakness of the serum BDG assay: its low specificity. The pooled specificities between the three meta-analyses ranged between 75 and 86%. Importantly, two of these meta-analyses excluded patients with known invasive fungal infection from the analysis, increasing the specificity for PJP. One of the meta-analyses also found that specificity declined to 73% when only non-HIV-infected patients were included (7).
Studies not included in these meta-analyses further expose the low specificity of BDG testing for a diagnosis of PJP. A study from a tertiary-care center in South Korea tested available patient and volunteer blood samples for the presence of BDG, including samples from 50 patients with documented PJP, and demonstrated a specificity of 74% using the recommended upper limit of the normal cutoff value of 80 pg/ml (8). An AIDS Clinical Trials Group study of patients presenting with respiratory symptoms found that that the specificity of this test for diagnosing PJP was 75% (9).
These consistently low diagnostic specificities should certainly raise concerns about the diagnostic value of the serum BDG test for diagnosing PJP. Use of this test alone undoubtedly leads to many false-positive results and has been shown to trigger inappropriate antibiotic use (1). Ascribing an elevated BDG level to PJP may also lead to further progression of a misdiagnosed and untreated process, often in patients who are immunocompromised and/or have critical illness.
Many infectious and noninfectious processes increase BDG levels. Systemic infections by almost all clinically relevant fungal organisms, except Zygomycetes and, in most cases, Blastomyces and Cryptococcus, are associated with elevated levels of BDG, as it is a component of the cell wall of many fungi (10, 11). It is for this reason that BDG is colloquially referred to as a “pan-fungal marker.”
Interestingly, elevated BDG levels are not exclusively seen in infections due to fungal organisms; they have also been demonstrated in patients with Streptococcus pneumoniae, Alcaligenes faecalis, and Pseudomonas bacteremia who were without other known causes of BDG assay reactivity (3, 12).
Several products and procedures may also increase serum BDG levels without any underlying infection. Examples include exposure of serosal surfaces to certain types of surgical gauze, hemodialysis performed with cellulose membranes, and the administration of intravenous immunoglobulin, albumin, and other blood products exposed to cellulose (3, 13–17). Since many of these exposures occur in patients who are at risk for both PJP and other systemic fungal infections, they are the source of considerable diagnostic uncertainty for clinicians.
Ironically, certain antibiotics test positive for BDG when diluted or solubilized to reconstituted-vial concentrations. Examples include colistin, ertapenem, cefazolin, trimethoprim-sulfamethoxazole, cefotaxime, cefepime, and ampicillin-sulbactam. This may not be clinically relevant, as no reactivity of the BDG assay was demonstrated when these antibiotics were diluted to the usual maximum plasma concentrations (18). Findings from a subsequent prospective study do, however, suggest that false-positive serum BDG assay results may be produced in patients receiving standard therapeutic doses of ampicillin-sulbactam (19). Thirty-seven of 117 serum samples from 15 patients receiving ampicillin-sulbactam in this study had a positive serum BDG assay. None of the 15 patients had invasive fungal infection, recent hemodialysis using cellulose membranes, or administration of intravenous albumin, immunoglobulin, or other products prepared using cellulose depth filters. Five of the 15 patients had undergone surgical debridement of a diabetic foot infection, though only 6 of 40 serum samples from these patients had serum BDG levels of >80 pg/ml (19).
Given the multitude of causes beyond PJP that can lead to elevations in serum BDG, it is not surprising that a single center cross-sectional study of 46 intensive care unit (ICU) patients revealed no significant difference in serum BDG levels between ICU patients with confirmed fungal infections and those with bacterial infections (20). A positive BDG test result in a critically ill patient with potential infection by, or exposure to, any of the above-named organisms or factors should therefore be interpreted with caution.
It has, however, been our experience that it is in these types of clinical scenarios, namely, immunocompromised patients with critical illness, in which serum BDG testing is commonly done: the patient with a recent solid organ transplant and a complicated postoperative course who develops acute respiratory distress syndrome, the patient receiving treatment for a hematologic malignancy who develops septic physiology and shows abnormalities on a CT scan of the chest, the critically ill patient with hepato-renal syndrome who is dialysis dependent and develops fever and hypoxia and is found after imaging to have bilateral pulmonary opacities, and the patient receiving immunosuppressive therapy for connective tissue disease who develops flu-like symptoms and altered mental status and is found to have imaging consistent with atypical pneumonia. Each of these clinical scenarios has a broad differential diagnosis which includes PJP. Each scenario also lends itself to the possibility that the patient may have recently received treatments or acquired an infection by other organisms which contain BDG. It is therefore inappropriate to rely solely on a positive serum BDG test result to make the diagnosis of PJP.
There are, however, situations in which a serum BDG test may be useful when considering a diagnosis of PJP. The high sensitivity that has previously been consistently demonstrated for this assay renders it an effective means of screening for PJP in the appropriate clinical setting, i.e., with patients who are not critically ill and in whom there is low-to-moderate suspicion for PJP. A negative serum BDG test in this setting would strongly support a diagnosis other than PJP. It should, however, be noted, that a recently published retrospective analysis of the serum BDG assay’s performance with over 400 hospitalized patients with malignancy and unexplained lung infiltrates revealed a sensitivity of only 69.8% when the manufacturer’s threshold of >80 pg/ml was used and when PJP PCR performed on bronchoalveolar lavage fluid, bronchial washings, or lung tissue was employed as the reference method (21). These findings suggest that the assay should not be used as a stand-alone means for ruling out PJP in immunocompromised patients with pneumonia.
Prior studies have demonstrated that many patients with PJP have a serum BDG level that is well above the standard cutoff of the commonly used assays. Koo et al. found that 71% of 14 patients with probable PJP had a serum BDG level of >500 pg/ml (3). Sax et al. found that 45% of 173 study participants with HIV and PJP had a serum BDG level of >500 pg/ml (4). An argument may therefore be made for utilizing serum BDG assays with higher cutoff values for diagnosis in select patients when there is both strong clinical suspicion and an appropriate level of pretest probability for PJP. It should, however, be noted that between 14 and 36% of patients in these two studies with serum BDG levels of >500 pg/ml did not have strong evidence of PJP, which suggests that the assay continues to lack the appropriate level of specificity even when the cutoff level for diagnosis is significantly increased. It has, however, recently been demonstrated that the specificity of the serum BDG test can be optimized to 100% in patients with a clinical picture consistent with PJP when the BDG level is >200 pg/ml and is followed up with a positive PJP PCR test (21).
Finally, the serum BDG assay itself may also have an important use as an adjunct test when used in tandem with P. jirovecii qualitative PCR (qPCR). Damiani et al. found that concurrent use of both low and high P. jirovecii qPCR cutoff values for bronchoalveolar lavage (BAL) samples resulted in this method achieving both 100% sensitivity and 100% specificity in the diagnosis of PJP. They also found that the serum BDG level was useful in determining whether indeterminate levels of P. jirovecii qPCR—those that fell between the high and low cutoff levels—in BAL fluid samples represented airway colonization versus true PJP (22).
In conclusion, the serum BDG assay lacks the specificity needed to make a diagnosis of PJP with this test alone. Positive serum BDG assays should be interpreted with caution, especially for critically ill patients with possible exposure to, or infection by, the many known additional causes of elevated BDG levels. Except in settings with a high pretest probability of PJP and a BDG level of >500 pg/ml, clinicians should not rely on serum BDG to make the diagnosis of PJP.
Edward F. Pilkington III and Paul E. Sax
