Introduction

It is estimated that 150,000 new infections due to coccidioidomycosis occur in the US each year. Most of these are acquired in the coccidioidal endemic regions, which include the southern San Joaquin Valley and other areas of California, much of Arizona, and portions of Nevada, Utah, and Texas [1]. In 40 % of these cases, infection manifests as a focal bronchopneumonia that can be difficult to distinguish from a community-acquired bacterial infection [2, 3]. The vast majority of these coccidioidal pneumonias resolve, but a frequent event is the development of a pulmonary nodule visible on a plain chest radiograph [4]. Coccidioidomycosis has been estimated to be the etiology of up to 25 % of pulmonary nodules encountered in the coccidioidal endemic area [5]. While the diagnosis of a pulmonary coccidioidal nodule may be clear when there is an initial diagnosis of acute pulmonary coccidioidomycosis and the radiologic evolution of the nodule has been documented [6], the diagnosis can be more difficult in other cases. Moreover, while there may be clues pointing to a diagnosis of coccidioidomycosis based on clinical, serologic, and radiographic findings [7–9], these cannot be completely relied upon to preclude a malignancy.

Because of this, other modalities have been employed in an attempt to distinguish coccidioidal nodules from malignant etiologies. By detecting the uptake of 18fluorine-labeled deoxyglucose (18FDG), positron emission tomography combined with thoracic computed tomography (PET/CT) has been used to assess for metabolically active lesions, including malignancies [10–14]. While inflammatory processes may be detected by this technique [15], PET/CT has been increasingly used to differentiate malignant from benign infectious nodules [11, 13–15]. We have observed several patients with pulmonary nodules that were due to coccidioidomycosis whose lesions were found to be positive using PET/CT. This led us to hypothesize that PET/CT frequently may be positive for pulmonary nodules due to coccidioidomycosis. To examine this further, we reviewed the results of PET/CT among a group of patients living in a coccidioidal endemic region who had undergone biopsy for a lung nodule that was found to be due to coccidioidomycosis or was granulomatous without an etiology, with the presumption that these were also due to coccidioidomycosis, and compared the results to those of patients in whom the pulmonary nodule was found to be due to a malignancy.

Methods

Patient Population

We retrospectively reviewed the records of all patients who had a diagnostic pulmonary biopsy of a pulmonary nodule at the Southern Arizona Veterans Affairs Health Care System (SAVAHCS) between January 2008 and March 2012 and identified those patients whose biopsies were either diagnostic for coccidioidomycosis, based on the identification of spherules, or were granulomatous but without a specific etiology. These biopsies were compared to those with a diagnosis of malignancy. A chart review was done with further identification of those individuals who had PET/CT as part of their diagnostic workup prior to performing the biopsy. The project was approved by the Human Subjects Protection Program of the Research and Development Committee of SAVAHCS (#13-3014). Informed consent was waived for the study.

Biopsy Assessment

All biopsied samples were subjected to standard laboratory specimen-handling processes in concordance with the College of American Pathologists and Joint Commission for Accreditation of HealthCare Organizations. For histopathological and cytopathological analyses, the biopsy materials and stained slides were examined by an experienced pathologist (MAR). The diagnosis of coccidioidomycosis was based on observing spherules in the biopsy specimen or growing Coccidioides from the specimen.

PET/CT Acquisition and Interpretation

PET/CT was performed on a Siemens Biograph 64 scanner. The CT scan was performed for attenuation correction as well as anatomic localization of radiotracer abnormalities. Patients were asked to fast for a minimum of 6 h; they were subsequently intravenously injected with 370 MBq (10 mCi) of 18FDG, followed within 60 min by PET/CT. Images were obtained from the orbits to mid-thigh level. Results were reported as positive or negative based on maximum standard uptake values that are based on body weight (SUVmax). All PET/CT scans were reviewed by an experienced nuclear medicine physician (TKW). A nodule or other area was considered positive if there was increased uptake of 18FDG compared to the blood pool [16].

Data Analysis

Data were analyzed using Stata® 13 (StataCorp, College Station, TX). Continuous variables were compared using the Mann-Whitney rank sum test while the χ2 test was used for discrete variables. A P value of < 0.05 was considered significant. Multivariable analysis was performed using factors with a P value < 0.10 on univariate analysis and an odds ratio (OR) with 95 % confidence interval (CI) was calculated from this. The nonparametric Kruskal-Wallis test was used in an analysis of variance to ascertain if there were differences among the various malignancies with regard to SUVmax or size.

Results

Demographics and Clinical Description of the Cohorts

A total of 351 cases had a diagnostic pulmonary nodule biopsy performed at SAVAHCS between January 1, 2008 and March 31, 2012. Among these, 51 (14.5 %) were either coccidioidal (n = 39) or granulomatous without a diagnosis (n = 12). The other 300 nodules were due to malignancy. Two patients had separate biopsies for different pulmonary lesions. The median age of those with coccidioidomycosis or a granulomatous process was 63 years compared to 69 years for those with cancer (P < 0.001). Two of the 51 with coccidioidomycosis or a granulomatous process were female compared to 11 of 300 with cancer (P = 0.929). Eight of the 51 subjects who were found to have either coccidioidomycosis or a granulomatous process on biopsy smoked tobacco compared to 24 with a cancer found on biopsy (P = 0.078). A coccidioidal serology was performed in 37 (72.5 %) of the 51 patients with a granulomatous process compared to 153 (51.0 %) of the 300 patients with malignancy (P < 0.001). A positive coccidioidal serology was found in 9 (24 %) of the 37 with a granulomatous process who had serology performed compared to 3 (2.0 %) of those with malignancies (P < 0.001).

Evaluation with PET/CT

A total of 245 (69.8 %) of the 351 who had a diagnostic biopsy had a prior PET/CT performed, including 15 (6.1 %) with a coccidioidal (n = 12) or a granulomatous process (n = 3) and 230 (93.9 %) with cancer. Comparison of these two groups is shown in Table 1. As can be seen, there was a trend toward a younger age in those with a granulomatous process and both the SUVmax and the nodule diameter were significantly lower in those with a granulomatous process compared to those with a malignancy (for both, P < 0.01). Coccidioidal serology was positive in 1 (8.3 %) of 12 with a granulomatous process compared with 2 (1.7 %) of 117 with malignancy (P = 0.147). The cohorts were overwhelmingly male and most of the subjects in both groups used tobacco products (P = 0.359). Uptake of 18FDG by lymph nodes in the mediastinum and hila was also assessed. Two (13.3 %) of the 15 subjects with granulomatous nodules had 18FDG uptake in these areas compared with 108 (47.0 %) of 122 with malignancies (P = 0.011). Multivariable analysis of age, nodule diameter, whether there was 18FDG uptake in the mediastinum or hila, and pulmonary nodule SUVmax revealed that only pulmonary nodule SUVmax was significantly different between those with malignancies and those with granulomatous lesions (OR 1.28, 95 % CI 1.05–1.55; P = 0.013).

Table 1 Comparison of 15 patients with a PET/CT scan and diagnostic biopsy of granulomatous pulmonary nodules to 230 with a biopsy diagnosis of malignancy
Full size table

The pulmonary nodule SUVmax for the 15 subjects with granulomatous lesions ranged from 0 in four subjects to a maximum of 5.9 for one subject. For malignancies, the SUVmax ranged from 0 in seven subjects to a value of 30. Figure 1 demonstrates the distribution of SUVmax among the subjects with granulomatous lesions and those with malignancies. While there is overlap at the lower SUVmax values, all nodules with SUVmax > 5.9 were due to malignancies. The sensitivity for distinguishing a malignant from a granulomatous coccidioidal process based on an SUVmax of >2.5, a value noted by Sim et al. [17] to be associated with malignancy, was 93 % and the specificity was 47 %. The positive predictive value was 96 % and the negative predictive value was 29 %. Using an SUVmax of >5.9, the highest value for a granulomatous coccidioidal lesion in this study, the sensitivity was 69 %, the specificity was 100 %, the positive predictive value was 100 %, and the negative predictive value was 17 %.

Fig. 1
figure 1

Vertical dot plot of the pulmonary nodule SUVmax of the 15 subjects with a coccidioidal or granulomatous etiology (Coccidioidomycosis) and that of the 230 subjects with malignancies (Cancer). Horizontal bars indicate median SUVmax for each group

Full size image

Analysis of the SUVmax among different types of malignancies was performed and is shown in Table 2. As can be seen, there is great overlap between both the SUVmax and the size of the nodules for the different groups of cancers. Nonparametric analysis of variance demonstrated no significant differences between these cancers with regard to either parameter (for both, P > 0.600).

Table 2 Comparison of pulmonary nodule SUVmax and size among the 230 malignancies
Full size table

Discussion

The noninvasive diagnosis of pulmonary nodules remains challenging [14, 15, 18] and no single method has proven to be entirely satisfactory. Several studies have suggested specificity for malignancy based on the detection of the uptake of 18FDG by PET either in the lesion or in the mediastinum. However, none of these studies was done in a coccidioidal endemic region. In the current study, we chose to study patients with pulmonary nodules that were biopsy-confirmed to be due to a granulomatous process likely to be coccidioidomycosis and compare them to subjects who were ultimately found to have a malignancy on biopsy. In the present study, all patients underwent pulmonary biopsy because of a presumed high risk of malignancy based on clinical parameters. None had evidence of clinically active coccidioidomycosis at the time of study.

Nearly 15 % of all the biopsies of pulmonary nodules in this study were diagnosed as granulomatous and 11 % were definitively coccidioidal. This percentage dropped in those who underwent PET/CT and is lower than noted in previous studies [5, 19–21]. These results still indicate that pulmonary nodules are not uncommonly found to be granulomatous in the coccidioidal endemic region. There have been two prior case reports of positive PET/CT associated with pulmonary coccidioidomycosis [22, 23]. Their findings of 18FDG uptake are similar to ours and indicate that our data are not anomalous.

We found several potentially distinguishing characteristics in subjects with pulmonary nodules due to a granulomatous process compared to malignancies, including older age, nodule size, uptake of 18FDG in the mediastinum or hila, and the SUVmax of the nodule. However, the only significant distinguishing characteristic was the SUVmax.

Weaknesses of this study include the relatively small number of subjects and possible bias given its retrospective, nonrandomized design. Unlike Gupta et al. [11], we did not assess the uptake of 18FDG over time nor were isotropic volume data generated on CT, as reported by Ohno et al. [24]. We were not able to ascertain the SUVmax at two separate time points, as reported by Kim et al. [25]. These methods may have further distinguished granulomatous from malignant lesions, particularly for lesions with low SUVmax. Despite this, the results indicate that pulmonary nodules due to coccidioidomycosis are frequently positive by PET/CT, but that a high SUVmax appears to distinguish a pulmonary nodule due to a malignancy from one due to coccidioidomycosis. These results are bolstered by three recent studies comparing PET results from pathologically benign and malignant pulmonary lesions [17, 26, 27]. In all three studies, inflammatory lesions, none of which were coccidioidal, had significantly lower SUVmax than those due to pulmonary malignancies, with values in both groups similar to what was observed here. In our study, all nodules with a SUVmax > 5.9 were malignant. However, given the small numbers, it is possible that some coccidioidal nodules could have a value exceeding this number. Moreover, most malignancies in this study had SUVmax > 2.5.

Several factors, such as age, smoking history, and whether the patient has a positive coccidioidal serologic test, should be considered when determining whether a pulmonary nodule in a patient living in a coccidioidal endemic region should undergo immediate biopsy or be followed clinically. The results of this study suggest that if PET/CT is performed, the results are most useful when the SUVmax is elevated above 5.9, suggesting that the lesion is malignant.