An Accuracy-Based Approach to the Microbiologic Diagnosis of Pulmonary Infection: Part III

Introduction: Similar to that of bacterial infection as described in Part 1 and fungal infection described in Part 2, the performance of diagnostic tests for viral pneumonia and parasitic pneumonia are not well described. Methods: We undertook a literature search to assess the accuracy of diagnostic tests for pneumonia, identified through a search of MEDLINE-indexed journals. Sensitivity and specificity of diagnostic tests for pneumonia were calculated with respect to various reference standards. Results: A battery of diagnostic testing is adequate to rule out most pathogens leading to viral pneumonia, lymphatic filariasis, and Toxoplasma. Testing is inadequate to exclude, and empiric treatment should be considered, for clinical suspicion of Hantavirus , Herpes Simplex Virus , Strongyloides , Roundworm, Hookworm, Paragonimus , and Toxocara . Conclusion: Most viral pathogens may be excluded using a combination of viral PCR and serology. In contrast, the presence of parasitic pathogens is difficult to exclude by current diagnostic testing. Clinical judgment is necessary in ruling out these causes of pneumonia.

use of multiplex PCR has increased viral detection, but gaps in the diagnosis of viral infection remain. In regions that are not endemic to parasitic infection, the yield of diagnostic testing is likewise uncertain.
In Part 3 of this review, we will assess the literature and discuss the accuracy of diagnostic tests for infectious pneumonia caused by both viral and parasitic pathogens.

Viral Pneumonia Influenza
Influenza virus is a single-stranded negative-sense RNA virus in the family Orthomyxoviridae consisting of two common subtypes, A and B.
The sensitivity of cell culture for Influenza A (Flu A) and Influenza B (Flu B) among PCR-positive cases has been reported at 69% [1] and 89% [2], and 81% [1], respectively, and reduced to 73% [3] in cases diagnosed by either serology or PCR. Culture is not ordinarily recommended, however, as it provides only 1% [3] additional sensitivity to the combination of PCR and serology, and requires over 72 hours, limiting its utility.

Introduction
Increased use of immunosuppressive medications, along with improved longevity of patients with oncologic diseases have led to increased complexity in diagnosing opportunistic infection of the lung. The RSV  A serologic diagnosis of Influenza can be made, but like culture, is less accurate than PCR. ELISA IgM sensitivity has been reported in 84% [3] of cases detected by either PCR or culture. A fourfold titer rise of EIA has been demonstrated in only 47% [6] of PCRpositive cases, but has been reported to provide an additional sensitivity of up to 34% [6] to that of PCR, and may be complementary. Similarly, among PCR-positive cases, IF has been reported positive in 71% [7], with a specificity of 94% [7], but IF requires serial evaluation to demonstrate a diagnostic fourfold titer rise. The time required to make a serologic diagnosis limits its utility, although the reduced specificity of serology illustrates the many PCR-negative cases. Both ELISA and CF are commercially available, but are not generally recommended for diagnosis.
PCR can be used to rapidly detect the Influenza virus. The pooled sensitivity of rapid NAAT for FluA and FluB have been reported 92% [5] and 95% [5], respectively, with a specificity of 98% [5]. As previously noted, there are cases that PCR fails to detect, but the sensitivity of PCR in cases detected by either culture or serology has been reported at 92% [3] and 98% [8], with a specificity of 84% [3] and 98% [8]. Viral mPCR PPA with Flu PCR has been reported at 90% [9], 93% [10], 95% [11], and 100% [12,13] with a NPA of 100% [9,11-13]. PCR remains the diagnostic test of choice for Flu due to its excellent diagnostic accuracy, but either empiric treatment or further diagnosis with culture and serology should be considered if there is ongoing clinical suspicion and the need for a diagnosis.

Parainfluenza virus
Parainfluenza virus (PIV) is a single-stranded negative-sense RNA virus in the family Paramyxoviridae made up of subtypes 1-4.
The sensitivity of PIV antigen detection by IF has been reported at 50% [17] and 53% [16] of PCR-positive cases, but as is true with culture, adds minimal additional sensitivity to PCR alone [16,17]. Likewise, conventional fluorescent antibody (FA) fails to add cases to those detected by PCR, with a sensitivity reported at 33% [18] and 66% [19], a rate which remains consistently a NPA of 98% [47] and 100% [9,11,12]. The reduced sensitivity against a DFA or culture reference standard indicates that multiple modalities may be necessary to confidently rule out infection.

Human enterovirus
Human Enteroviruses (EVs) are positive-sense RNA viruses in the family Picornaviridae that can lead to either upper or lower respiratory tract infection. Cell culture sensitivity has been reported at 41% [74] and 50% [75] of PCR-positive cases. Culture provides additional sensitivity to the use of PCR alone, but like culture for most viral infections, requires time and skill and should only be considered when PCR fails to detect the virus.
BALF DFA sensitivity has been reported at 80% [76] of PCR-positive cases, but is not recommended due to the superior sensitivity of PCR. Additionally, it is by no means certain that it provides additional sensitivity to PCR.
Among cases of SARS-CoV-2 PCR-positive infection, chemiluminescent immunoassay (CLIA) IgM or IgG pooled sensitivity has been reported at 98% [77], with a specificity of 97% to 98% [77]. ELISA IgM or IgG pooled sensitivity has been reported at 84% [77] of PCR-positive cases, with a specificity of 98% [77]. LFIA IgM or IgG pooled sensitivity has been reported at 66% [77] of PCR-positive cases, with a specificity of 97% [77]. Among PCR-positive cases, the pooled sensitivity of either CLIA or ELISA has been reported at 85% [78,79], with a specificity of 92% [79] and 99% [78]. CLIA serology can provide additional cases to PCR and should be considered when PCR is negative.
Among cases of SARS-CoV-1 infection, loop-mediated amplification (LAMP) sensitivity has been reported at 71% [80] of cases detected by either serology or PCR, but it is not widely commercially available.
Nasopharyngeal PCR sensitivity for 229E has been reported at 50% [81] of culture-positive cases and 62% [76] of cases positive by either culture or serology. PCR for OC43 has been reported at 40% [81] of culturepositive cases and 100% [76] of cases positive by either culture or serology. PCR for HK1 has been reported at 100% [76] of cases positive by either culture or serology. PCR for NL63 has also been reported at 100% cases. In addition to its limited diagnostic value, serology is not commercially available. The pp65 antigen has been reported to add up to an additional pooled 12% [100] sensitivity to the use of PCR alone. Therefore, PCR and pp65 antigen testing should ideally be used concurrently, although the pp65 antigen is not widely commercially available.
EIA IgG serology sensitivity has been reported at 30% [101] of PCR-positive cases, but has poor specificity due to cross-reaction with Epstein-Barr Virus and Human herpesvirus-6. IgM sensitivity has been reported as [76] of cases detected by either culture or serology. The sensitivity of PCR for these four strains combined has been reported at 87% [76] of cases positive by either culture or serology. The sensitivity of PCR for SARS-CoV-1 has been reported at 78% [80] of serologicallydetected cases, while PCR sensitivity for MERS-CoV has been reported at 100% [82] of culture-positive cases, with a specificity of 100% [82], but evaluated in only a small sample. SARS-CoV-2 nasal PCR sensitivity has been reported at 73% [80] of cases positive by either culture or serology. The accuracy can vary by the source of collection with sputum, saliva, and nasopharyngeal aspirate exhibiting a pooled sensitivity of 97% [79], 62% [79], and 73% [79]
HBoV has not yet been isolated in cell culture and thus the sensitivity of viral culture is currently unknown.
HBoV-1 antigen sensitivity has been reported at 76% [87] of PCR-positive cases, with a specificity of 100% [87]. Antigen testing is also not commercially available, but also would not appear to add any increased sensitivity over PCR.

Varicella-zoster virus (VZV)
Varicella-Zoster Virus (VZV) is a member of the family Herpesviridae, which may lead to pneumonia in immunocompromised patients. Very little data is available regarding the diagnosis of Varicella pneumonia, and the diagnosis is generally made from the characteristic skin rash.
The sensitivity of culture for vesicular lesions has been reported at only 44% [107] of clinically diagnosed cases and 56% [115] of DFA-positive cases, with a specificity of 97% [115]. Culture may be performed on BALF, although like HSV it has been poorly studied and the sensitivity of respiratory secretion culture is unknown. Presumably its sensitivity will be less than that of vesicular lesion culture.
DFA sensitivity for vesicular lesions has been reported at 92% [115] of culture-positive cases, with a specificity of 80% [115]. DFA can therefore detect additional cases to the poorly sensitive culture but has not yet been well studied in BALF and so has uncertain utility.
Serological methods may be utilized, but only with limited accuracy. ELISA IgM sensitivity has been reported at 25% [116] of PCR-positive cases. Like other diagnostic tests for VZV, serology has not been well-studied in cases of pneumonia. Both EIA and CLIA serology are commercially available, but there as yet is no clear role for serology in the diagnosis of VZV pneumonia.

Hantavirus
Hantaviruses are single-stranded, negative-sense RNA viruses of the family Bunyaviridae, consisting of over 20 different strains [118].
Culture is not generally performed due to the risk it presents to laboratory personnel, and moreover the sensitivity is uncertain.
ELISA IgM sensitivity has been reported at 73% [118] and 97% [119] of clinically diagnosed cases, and 100% [120] of PCR-positive cases, with a specificity of 100% [119]. PCR is more likely than ELISA to be positive in the first seven days, whereas ELISA is more likely to be positive thereafter. IF [136]. The specificity has been reported at 82% [136], and 87% to 100% [138]. CF sensitivity has been reported at 89% [135] of those found positive by ELISA IgM and similarly at 89% [130] of those positive by IF. CF is not commercially available, however, and also requires a longer processing time than ELISA, limiting its utility. Lastly, CLIA sensitivity has been reported at 97% [139] of cases diagnosed by ELISA IgM with a specificity of 93% [139]. Any one of the three available techniques (IF, ELISA, or CLIA) may be positive when used alone, and therefore either combining methods or using them sequentially can increase sensitivity.
The sensitivity of PCR from respiratory secretions has been reported at 53% [140], 90% [133], and 98% [141] of seropositive cases by ELISA, 68% [133], 81% [142], and 96% [137] of clinically diagnosed cases, 85% [143] of culture-positive cases, and at 90% [143] of LAMPpositive cases. The sensitivity of PCR has been reported to be more than 50% higher [140] in the first two weeks of the rash than in the following two weeks and is thus the preferred test in the early stage of infection. PCR appears to have a higher sensitivity than serology in the acute phase, while serology sensitivity is superior in the convalescent phase. PCR has not been well evaluated in BALF, but can be performed in cases in which there is clinical suspicion.
LAMP sensitivity of respiratory secretions has been reported at 100% [143] of PCR-positive cases, and 95% [143] of culture-positive cases, but is not commercially available and also has uncertain utility ( Figure 1).

Parasitic Pneumonia Lymphatic filariasis (Wuchereria bancrofti, Brugia malayi, Brugia timori, and others)
Lymphatic filariasis is a parasitic infection caused by nematodes such as Wuchereria bancrofti, Brugia malayi, and Brugia timori which are is transmitted to by ELISA, with a specificity approaching 100% [122]. ELISA serology is commercially available and should be used for the initial diagnosis of Hantavirus.

Measles
Measles virus is a negative-sense RNA virus in the family Paramyxoviridae that can lead to pneumonia, bronchiolitis, or bronchitis.
Various serologic techniques can be used to diagnose Measles, including commercially available IF, ELISA, and CLIA. The sensitivity of IF has been reported at 34% [132] of PCR-positive cases, 46% [127] and 83% [130] of clinically diagnosed cases, and at 68% [127] and 89% [129] of culture-positive cases. IF and culture can be complementary as either may be individually positive. ELISA IgM sensitivity has been reported at 56% [133] of PCR-positive cases, and at 63% [134], 74% [135,136] 85% [137], and 98% to 100% [138] of clinically diagnosed cases. In the acute phase, IgM has been reported at 58% to 85% [138] of CF-positive cases, but in the convalescent phase has been reported to increase to 88% [136], 93% to 98% [138] and 97% [135], and may remain positive for over one month from the

strongyloides stercoralis
Strongyloides stercoralis is a nematode transmitted directly to humans through soil, leading to pneumonia with eosinophilia.
The sensitivity of the stool microscopic examination varies according to the method employed. Among clinically diagnosed cases, the pooled sensitivity of the Baermann method has been reported at 72% [162], agar plate culture (APC) at 89% [162], direct examination at 21% [162], and formaldehyde-ether (FEC) pooled at 48% [162]. Obtaining serial stool cultures can increase sensitivity. For example, among culture positive cases tested over seven consecutive days, only 53% [163] have been reported positive on day one. Stool APC appears to be the preferred coprologic test and among cases detected by either microscopy, FEC, Harada-Mori filter paper culture, or APC, the sensitivity has been reported at 96% [164]. Among cases detected by PCR, however, the sensitivity has been reported at only 18% [165]. Of these PCR-positive cases, the addition of the Baermann method to APC has been reported to increase sensitivity by 24% [165]. The reduced sensitivity when PCR is added to the reference standard indicates that coprologic techniques alone are insufficient to make a trustworthy diagnosis. Among cases serologically confirmed by ELISA or LIPS, the sensitivity of microscopy by either FEC, Baermann, or stool APC has been reported at 53% [166]. Lastly, of serologically diagnosed cases by IF, the sensitivity of microscopy by Lutz, Rugai, or APC has been reported at 95% [167]. Microscopic examination can be performed on BALF, although it is not certain to provide additional sensitivity to fecal microscopy [168].
Stool PCR can be used to detect Strongyloides. The pooled sensitivity of PCR has been reported at 71% [172] of cases detected by microscopy but reduced to 57% [172] of cases detected by either serology or microscopy, with a specificity of 95% [172]. Because serology may sometimes be positive due to remote infection, the true sensitivity is on the upper end of the range, but when used in isolation it lacks the sensitivity to rule out active infection. PCR on BALF is neither well studied nor commercially available.
humans through a mosquito vector, leading to tropical pulmonary eosinophilia.
CFA cards can be used to detect the Onchocerca gibsoni circulating antigen (Og4C3) of Wuchereria. ELISA antigen sensitivity for Wuchereria has been reported at 95% [148][149][150], 98% [151], 99% [152], and 100% [145,146,153,154] of cases detected by either mf micropore or thick smear, with a specificity of 94% [153], 98% [151], and 100% [149,150], indicating the technique's ability to identify mf-negative cases. ELISA CFA sensitivity for Brugia is slightly less, having been reported at 75% [144] and 83% [149] of cases positive by microscopic blood smear. The CFA immunochromatographic test (ICT) sensitivity among serology-positive cases for Wuchereria has been reported at 90% [155], with a specificity approaching 100% [155]. The sensitivity, however, decreases to 60% [155] following the administration of drugs which are commonly used in endemic regions. In low-prevalence areas with cases which have been detected by the ELISA CFA, ICT sensitivity has been reported at only 36% [156]. Although it is not widely commercially available, ELISA CFA assay is recommended in conjunction with microscopy.
Serum PCR sensitivity for Wuchereria has been reported at 94% [143] of clinically diagnosed cases and 100% [147,161] of mf-positive cases, with a specificity of 100% [161]. The sensitivity of serum PCR for Brugia is uncertain, however the specificity approaches 100% [161] of microscopically detected cases. Neither Wuchereria nor Brugia PCR is commercially available and both are uncertain to add any additional sensitivity over CFA. Furthermore, the accuracy of PCR on BALF has not yet been thoroughly evaluated. • Page 9 of 18 • 95% [179], 99% [173], and 100% [175,187].

Paragonimus (westermani, mexicanus, africanus, and others)
Paragonimus is a lung fluke from the Trematoda phylum that infects the lungs of humans who ingest crustaceans which are eaten raw or undercooked.
Paragonimus eggs can be identified by microscopy in either sputum or stool samples, as patients may unknowingly swallow sputum. The sensitivity of sputum microscopy has been reported at 10% [185] and 62% [188] of cases serologically detected by ELISA, and 14% [185] and 46% [189], and 72% [190] of clinically diagnosed cases. Of these clinically diagnosed cases of pulmonary paragonimiasis, the sensitivity of stool detection has been reported to reduce to 65% [190]. Either sputum or stool can be independently positive, and of clinically diagnosed cases the sensitivity of combining stool and sputum can increase sensitivity to 85% [190]. Therefore, detection by both sputum and stool microscopy is recommended concurrently for initial testing. Empiric treatment should be considered for microscopy-negative cases in which there is still clinical suspicion, as alternative diagnostic tests are not commercially available.
Skin testing is infrequently utilized, but has been reported positive in 97% [191] of cases detected by either stool or sputum microscopy, but is not widely commercially available.
Serum antigen sensitivity has been reported at 42% [192] and 86% [193] of clinically diagnosed cases, and at 100% [193,194] of microscopy-positive cases, with a specificity of 94% [192]. The paragonimiasis antigen is not commercially available and is not currently recommended.
PCR has been performed on BALF, but is neither well studied nor commercially available.

Toxocara (canis, cati, and others)
Toxocara is a parasitic, zoonotic roundworm transmitted by the fecal-oral route which manifests with allergic symptoms, neurologic symptoms, and pneumonia. Hosts of the roundworm include cats, dogs, foxes, coyotes, and wolves.
The sensitivity of culture is not known and thus BALF culture is generally not recommended.

Roundworm (ascaris lumbricoides, ascaris suum)
Roundworms are a parasite from the nematode phylum transmitted to humans through the fecaloral route. Pneumonia occurs as a result of migration through the lungs.
PCR has been used to detect Ascaris lumbricoides. Stool PCR sensitivity has been reported at 31% [173] of cases detected by the KK method, 82% [179] of those detected by the FEC method, and at 80% [178], 95% [175], 97% [174], and 100% [180], of all microcopypositive cases, with a specificity of 81% [173], and 83% [179], 93% [175], and 95% [174]. As is the case with other parasitic infections, the accuracy of PCR on BALF is unknown. Unfortunately, Ascaris PCR for either stool specimens or BALF is not widely commercially available. LAMP sensitivity has been reported at 96% [181] of cases detected by the microscopic thick smear technique, with a specificity of 62% [181], but is also not widely commercially available.

Hookworm (ancylostoma duodenale, necator americanus)
Hookworms are helminth nematode parasites transmitted through soil that following migration to the lungs, lead to eosinophilic pneumonia.
BALF PCR sensitivity has been reported at 100% [204] of cases positive by microscopy [204] but only 50% [206] of those with a serologic titer rise, although as yet has been reported in only small numbers. As PCR is not known to provide additional sensitivity to the combination of microscopy and serology, its use is not currently recommended [207]. PCR is commercially available and may be considered, however, in cases of high clinical suspicion with either equivocal serology or negative BALF culture (Figure 1 and Figure 2).

Conclusion
Although many pathogens still lack effective treatment, there may be value in identifying the presence of infection. Hospitalized patients with pneumonitis found to be PCR-and culture-negative are commonly diagnosed with drug-induced pneumonitis, radiation pneumonitis, organizing pneumonia, or nonspecific inflammation. In these instances, prolonged immunosuppressive therapy is commonly administered, which itself carries a risk of exacerbating the undiagnosed infection. Therefore, it is desirable to determine the causative organism to before treating with immunosuppressive therapies. Accurate diagnostic tests for several viral and parasitic pathogens are lacking. These instances require astute clinical judgement and a high index of suspicion is needed. Pathogen-targeted therapy should be considered even in the presence of non-diagnostic tests for most infectious agents.
Serology is an effective means of diagnosis. EIA IgG sensitivity among clinically diagnosed cases has been reported uniformly at 87% [199], 91% [200], 92% [201], with a specificity of 86% [200], and 100% [199], although specificity can be limited due to cross-reactivity with Strongyloides and Trichinella. The sensitivity of ELISA IgE with a titer greater than 1 TU/liter has been reported at 81% [202] of cases positive by Western Blot, with a specificity of 54% [202]. Above a higher threshold of 50 TU/liter, the sensitivity lowers to 42% [202] while specificity increases to 96% [202]. Conversely, among cases positive by ELISA, the sensitivity of Western blotting has been reported at 100% [202]. Only Toxocara ELISA is commercially available but if negative, empiric treatment should still be considered [203].
While PCR has been performed on animals, information on its accuracy when used on human tissues is lacking. In addition, PCR is not commercially available.

Toxoplasma gondii
Toxoplasma gondii is an obligate intracellular protozoan acquired through ingestion of uncooked meat or infected water. Risk factors include pregnancy, HIV, and solid organ transplantation.
The sensitivity of BALF culture is unknown but culture of respiratory secretions can be performed [204]. Tissue culture sensitivity, however, has been reported at 100% [205] of microscopy-positive or serology-positive cases. The sensitivity of culture in fetal specimens with serologic evidence of infection, however, has been reported at only 80% [206]. Consistent with these findings, microscopic detection of BALF organisms by a Giemsa stain has been reported at 75% [204] of autopsy-confirmed cases, but increases to as high as 100% [204,207] of PCR-positive cases. BALF IF sensitivity has been reported at 33% [204] and 100% [205] of microscopy-positive cases, although as yet it has been evaluated only in small numbers. Because serology is not commercially available, microscopynegative cases should be followed by either tissue sampling or empiric treatment.

Financial and Funding Disclosure
There is no financial support or funding to report.

Conflicts of Interest
There are no conflicts of interest to report by any author.