Clinical and Radiologic Factors Associated with Pulmonary Nodule Etiology in Organ Transplant Recipients
Presented in part at the World Transplant Congress, Boston, MA, July 2006.
*Corresponding author: Ajit P. Limaye, firstname.lastname@example.org
Identifying clinical and radiographic factors that are associated with a specific etiology of pulmonary nodules (PNs) in solid-organ transplant (SOT) recipients might be helpful in guiding empiric therapy. Multivariable logistic regression was used to assess the relationship of clinical and radiographic variables to the etiology of PN in a retrospectively identified cohort of SOT recipients at a single transplant center. PNs in 55 SOT recipients (lung 15%, heart 22%, liver 42%, kidney 18% or kidney/pancreas 5%) were diagnosed at a mean of 1061 days post-transplant and were infectious in 31 of 55 (56%) (bacterial 22%, fungal 33%, viral 2%) and noninfectious in 24 of 55 (44%) [post-transplant lymphoproliferative disorder (PTLD) 25%, carcinoma 18%]. Radiographic ‘consolidation’ was independently associated with an infectious etiology (OR, 20.2, p < 0.01). Epstein-Barr virus seronegativity and lung transplant were each associated with PTLD (OR, 21.7, p < 0.01) and (OR, 36.6, p < 0.001), respectively. Diagnosis less than 90 days post-transplant was associated with Aspergillus infection (OR, 12.9, p = 0.007). Specific clinical and radiographic features are associated with specific etiologies of PNs in SOT recipients and might be useful for guiding empiric therapy while awaiting results of definitive diagnostic studies.
A range of pulmonary complications have been described in solid-organ transplant (SOT) recipients (1–5). Pulmonary nodules (PNs), in particular, have previously been shown to be relatively common in this setting (6–8), and pose a considerable clinical challenge given the broad differential diagnosis they represent (8–14). An aggressive diagnostic evaluation is usually undertaken given the serious nature of the common etiologies for PNs in this population, including various infectious and malignant processes. A definitive diagnosis might ultimately require multiple diagnostic procedures ranging from bronchoscopy with bronchioalveolar lavage to surgical biopsy and may take several days even in tertiary care settings at transplant centers. Until the results of definitive studies are available, empiric therapy is often given on the basis of available clinical and radiologic information. Information about the association of specific clinical, laboratory, or radiographic factors and the specific etiology of PNs in SOT recipients could be helpful in guiding empiric therapy while awaiting results of definitive diagnostic studies.
While previous studies have described the clinical and radiographic features of PNs in SOT recipients (8–14), they have been limited by one or more of the following: Lack of multivariable statistical analysis, small numbers of patients, or inclusion of only one or two types of SOT recipients.
Given these limitations of prior studies, we sought to examine the association between a range of clinical and radiographic factors and the etiology of PNs (as defined by histopathology and/or culture) in a relatively large cohort of SOT recipients with PNs using multivariate statistical analysis. We hypothesized that specific clinical and/or radiographic variables might be associated with specific etiologies of PNs in this setting, thereby providing useful guidance for selection of empiric therapy while awaiting results of definitive diagnostic studies.
A PubMed search was conducted in April 2006 using the keywords (‘pulmonary nodule**’ OR ‘coin lesion’ OR ‘lung nodule’) AND ‘transplant*’. Results were manually filtered, and related articles and pertinent references were reviewed. Specific criteria for inclusion included: Inclusion of 10 or more patients, documentation of 1 or more PNs, receipt of an organ transplant and description of the etiology of the PN(s).
Data collection and study design
We conducted a retrospective analysis of all SOT recipients who were diagnosed with PNs between 1990 and July 2005 at the University of Washington Medical Center. Initially, we identified all SOT recipients with PNs during this time period via a clinical database. We defined a PN as a roughly spherical or oval intrapulmonary radiographic opacity that was well enough circumscribed to permit measurement of its various dimensions and was limited to 6 cm or less in its largest dimension (15). We defined ‘consolidation’ as an area of increased attenuation of the lung of sufficient degree to obscure the lung vessels as previously described (16). A detailed summary of each of the specific patient and radiographic variables and specific outcomes analyzed are shown in Table 1. Each radiographic study (chest computed tomography scan) was analyzed for the characteristics of the PNs by a single attending chest radiologist (JDG) who was blinded to the ultimate etiology of the PN (Table 1). The final diagnosis of PN etiology was made on the basis of available pathologic, microbiologic and clinical data extracted from the electronic medical record using a standardized data collection form. The study was approved by the University of Washington Institutional Review Board (Human Subjects Division).
Table 1. Summary of variables and outcomes analyzed
| Patient characteristics||Age at transplant|
|Allograft rejection (within 30 days)|
|Cancer in explanted organ|
|Time of onset post-transplant|
| Radiographic characteristics||Size of largest nodule|
|Number of nodules|
|Outcomes analyzed||Infectious vs. noninfectious|
|PTLD vs. Non-PTLD|
|Aspergillus vs. non-Aspergillus|
The association of various clinical and radiologic characteristics with specific PN etiologies was determined. Three specific outcomes were chosen for analysis (infectious vs. non-infectious, Aspergillus vs. non-Aspergillus, PTLD vs. non-PTLD) on the basis of their established importance in SOT recipients, availability of a specific intervention that could potentially be initiated while awaiting definitive diagnosis, and inclusion of adequate number of cases for meaningful statistical analysis. Odds ratios (ORs) and confidence intervals (CIs) were estimated using logistic regression models. Risk factors that were significant at the 0.10 level in univariate models were included in multivariable models as appropriate. Specifically, variables were entered into a multivariable model when a minimum of 10 outcomes per variable were present (17). To estimate odds in EBV seronegative patients where cell counts were 0, exact logistic regression models were analyzed. All statistical analyses were run using SAS (version 8.0, SAS Institute Inc., Cary, NC, USA) or LogXact (LogXact-4, Cytel Software Corp., Cambridge, MA, USA) software.
During the study period, a total of 94 SOT recipients with possible PN were identified from a clinical database. Thirty-nine of these patients were excluded for the following reasons: No medical records available (n = 7), etiology of PN not determined (n = 11), imaging studies not available (n = 12), discrete PN(s) not confirmed on review of radiologic studies (n = 9), leaving 55 patients who met the inclusion criteria (history of organ transplant, presence of 1 or more PNs, medical records and imaging studies available for review) and were included in the analysis. The characteristics of the 55 patients who constituted the study population are shown in Table 2. The 28-day crude mortality rate after diagnosis of PNs was 3.6% (2 of 55).
Table 2. Characteristics of the study population
|Number of patients||55|
|Age at diagnosis (mean, range)||53.5 (33, 73)|
| Male||34 (62)|
| Female||21 (38)|
|Transplant type|| |
| Liver||23 (42)|
| Kidney||10 (18)|
| Kidney and pancreas||3 (5)|
| Lung||8 (15)|
| Heart||12 (22)|
|Recipient EBV serostatus|| |
| Negative||3 (5)|
| Positive||47 (85)|
| Unknown||5 (9)|
|Days between transplant and diagnosis|| |
| Range||16 to 10 050|
| 25th percentile||125|
| 75th percentile||1182|
|Rejection within 30 days|| |
| Rejection||10 (18%)|
| No rejection||39 (71%)|
| Unknown||6 (11%)|
|Cancer in explanted organ|| |
| Cancer||5 (9%)|
| No cancer||35 (64%)|
| Unknown||15 (27%)|
Table 3 shows the detailed radiographic characteristics extracted from the chest CT scan that demonstrated PNs.
Table 3. Radiographic characteristics of pulmonary nodules
|Mean size of largest nodule (range)||26 mm (6–60 mm)|
|Multiple nodules (≥2)||33 (60%)|
|Nodules bilateral||21 (38%)|
|Predominant nodule position|| |
| Central||14 (25)|
| Peripheral||28 (51%)|
| Both||11 (20%|
|Nodule border|| |
| Well-defined||46 (84%)|
| Ill-defined||8 (15%)|
| Spiculated||6 (11%)|
| Lobulated||6 (11%)|
|Other radiograph features|| |
| Cavitation||7 (13%)|
| Pleural effusion||19 (35%)|
| Consolidation||20 (35%)|
| Lymphadenopathy||12 (22%)|
Etiology of pulmonary nodules
Table 4 shows the etiology of the PNs as defined by histologic, clinical and/or microbiologic data from clinically performed diagnostic studies. Non-infectious etiologies accounted for slightly less than half of cases, with PTLD and carcinoma being the most common diagnoses. Among infectious etiologies, Nocardia spp. and Aspergillus were most common. A diagnosis was established by sputum culture in 1 (2%), bronchoscopy/bronchoalveolar lavage (BAL) in 9 (16%), transthoracic biopsy in 21 (38%) and surgical biopsy in 24 (44%).
Table 4. Etiology of pulmonary nodules
| Bacterial||12 (22)|
| Nocardia||6 (11)|
| Legionella||2 (4)|
| Mycobacteria||3 (6)|
| Other||1 (2)|
| Fungal||18 (33)|
| Aspergillus||11 (20)|
| Cryptococcus||4 (7)|
| Other||3 (6)|
| Viral||1 (2)|
| CMV||1 (2)|
| PTLD||14 (26)|
| Carcinoma||10 (18)|
| Hepatocellular||2 (4)|
| Non-small cell||3 (6)|
| Squamous cell||3 (6)|
| Unspecified||2 (4)|
Risk factors for specific pulmonary nodule etiology
Infection: Table 5 shows the factors associated with an infectious etiology. Receipt of a lung (vs. nonlung) transplant was independently associated with a noninfectious etiology of PNs in multivariate analysis [OR 15.95 (95% CI 1.21–209.6), p = 0.04]. In contrast, radiographic evidence of consolidation was strongly and independently associated with an infectious (rather than noninfectious) etiology [OR 20.15 (95% CI 2.57–940.7), p < 0.01].
Table 5. Risk factors for infectious etiology of pulmonary nodules
| Transplant||Other||29 (93.5)||18 (75)||4.8 (0.9–26.6)||0.07||20.7 (1.6–26.7)||0.02|
|Lung||2 (6.5)||6 (25)||1.0|| |
| Well-defined||No||8 (25.8)||1 (4.2)||8.0 (0.9–69.2)||0.06|| |
|Yes||23 (74.2)||23 (95.8)||1.0|| |
| Ill-defined||No||24 (77.4)||23 (95.8)||1.0|| |
|Yes||7 (22.6)||1 (4.2)||6.7 (0.8–58.9)||0.09|| |
| Consolidation||No||14 (45.2)||21 (87.5)||1.0|| ||1.0|| |
|Yes||17 (54.8)||3 (12.5)||8.5 (2.1–34.5)||0.003||20.2 (2.6–940.7)||<0.01|
PTLD: Table 6 shows the factors associated with PTLD as the etiology of PNs. Among the variables analyzed, only EBV seronegativity at transplant and receipt of a lung (vs. other organ) transplant were associated with PTLD in univariate analysis. Although the sensitivity of EBV seronegativity as a marker for PTLD etiology was low (21%), the specificity was 100%, but there were only three seronegative patients, all of whom were diagnosed with PTLD. The small number of outcomes precluded a multivariate analysis.
Table 6. Risk factors for post-transplant lymphoproliferative disorder (PTLD) etiology of pulmonary nodules
| Transplant||Lung||6 (42.9)||2 (4.9)||14.6 (2.5–86.0)||0.003|
|Other||8 (57.1)||39 (95.1)||1.0|| |
| EBV status at tx||Negative||3 (21.4)||0 (0.0)||11.3 (1.2–∞)||0.04|
|Positive||11 (78.6)||36 (87.8)||1.0|| |
| Location||Both||5 (35.7)||6 (14.6)||1.0|| |
|Peripheral||8 (57.1)||20 (48.8)||0.5 (0.1–2.0)||0.97|
|Central||1 (7.1)||13 (31.7)||0.1 (0.1–1.0)||0.05|
| Consolidation||No||12 (85.7)||23 (56.1)||1.0|| |
|Yes||2 (14.3)||18 (43.9)||0.2 (0.1–1.1)||0.06|
Aspergillus: Table 7 shows the factors associated with Aspergillus as the etiology of PNs. Only receipt of a heart (vs. non-heart) transplant and diagnosis of PNs within 90 days of transplant were each associated with Aspergillus (vs. non-Aspergillus) as the etiology in univariate analysis. The variable ‘diagnosis <90 days post-transplant’ had a sensitivity of 64% and specificity of nearly 89% for the diagnosis of Aspergillus. No specific radiologic characteristics were helpful in distinguishing Aspergillus from non-Aspergillus etiologies of PNs.
Table 7. Risk factors for Aspergillus etiology of pulmonary nodules
| Gender||Female||7 (63.6)||30 (68.2)||3.8 (0.9–15.0)||0.06|
|Male||4 (36.4)||14 (31.8)||1.0|| |
| Transplant||Heart||5 (45.5)||7 (15.9)||4.4 (1.1–18.5)||0.04|
|Other||6 (54.5)||37 (84.1)||1.0|| |
| Diagnosis post-tx (days)||≤90 days||7 (63.6)||5 (11.4)||13.7 (2.9–63.8)||<0.01|
|>90 days||4 (36.4)||39 (88.6)||1.0|| |
Sixty-three articles were identified using the search strategy described above and were individually reviewed. Among these screened articles, only six met the inclusion criteria and are summarized in Table 8.
Table 8. Previously published English language studies of pulmonary nodules in SOT recipients
|Journal||Chest||Radiology||J Thoracic Imaging||Medicine||J Heart Lung Transplant||Transplant||Am J Transplant|| |
|Number of subjects||23||25||15||33||13||33||55||197|
|Transplant type||Lung||Heart||Heart-lung, heart||Liver||Heart||Heart||Heart, lung, kidney, liver, kidney/pancreas|| |
| Aspergillus||35%||32%||25%||21%||38%|| ||20%||22%|
| Cryptococcus||0%||0%||0%||6%||0%|| ||7%||3%|
| Other||0%||0%||0%||15%||0%|| ||6%||4%|
| Viral – CMV||9%||7%||0%||0%||23%||6%||2%||5%|
| PTLD||39%||0%||0%||3%|| ||6%||26%||13%|
| Carcinoma||13%||0%||13%||21%|| ||12%||18%||13%|
| Other neoplasm||0%||7%||13%||0%|| ||6%||0%||3%|
| Other1||0%||11%||0%||12%|| ||27%|| ||8%|
|Mean onset after transplant (days)||603||141||330||249||N/A||N/A||1,060||588|
|Statistical analysis||No||No||No||No||No||No||Yes|| |
By means of a retrospective analysis of a large cohort of SOT recipients, we have demonstrated that several radiologic and patient characteristics are independently associated with a specific etiology of PNs. These findings have the potential for assisting in the selection of empiric therapy in the SOT recipient with PNs, at least until the results of definitive diagnostic studies become available. The major findings of this study were that the radiologic feature of ‘consolidation’ was strongly associated with infectious etiology regardless of organ transplanted and that EBV seronegativity and lung transplant (compared with other organ transplant type) were strongly associated with PTLD. In addition, PNs found early in the post-transplant period (less than 90 days) were much more likely to be due to Aspergillus than those that were diagnosed after 90 days. An unexpected finding was that radiographic features (nodule characteristics or size, distribution) were poorly correlated with the ultimate PN etiology.
What is the clinical relevance of these findings in the evaluation of an organ transplant recipient with PNs? In clinical practice, several days and diagnostic procedures might ultimately be required to make a specific diagnosis of PNs. While awaiting results of definitive diagnostic studies, the clinician may elect to begin empiric therapy on the basis of clinical impression, laboratory studies, etc. However, to our knowledge, a systematic evaluation of radiographic and clinical variables in predicting the etiology of PNs in this setting has not been reported. We found that among the few EBV seronegative patients, all had PTLD as the etiology of their PNs. Thus, among SOT recipients with PNs in this study, EBV seronegativity was highly predictive of PTLD (three of three patients). However, since EBV seronegativity is uncommon in adult SOT recipients, it would be important in future studies to assess whether this relationship is found in pediatric SOT recipients in whom EBV seronegativity is common. Furthermore, given the small number of patients, this interesting finding should be confirmed in other adult SOT populations.
Other findings in the study that might be useful clinically included the radiographic feature of consolidation. This finding had a sensitivity of 55% and specificity of 88% for an infectious (vs. non-infectious) etiology of PNs, and might therefore serve as a useful threshold for initiating empiric antimicrobial therapy in the appropriate clinical setting while awaiting results of definitive diagnostic studies. For Aspergillosis, the most useful factor was the timing of nodules post-transplant. Onset of PNs <90 days post-transplant had a sensitivity of 64% and specificity of 89% for Aspergillosis in this study. Thus, given the relatively high prevalence of Aspergillus in this setting, empiric antifungal therapy might be appropriate while the results of diagnostic studies are pending. Clearly, despite the strong and independent statistical association of timing of nodules and Aspergillus etiology (and the other variables described above), no single factor had a high-enough sensitivity or specificity or accuracy to ‘rule out’ or ‘rule in’ a specific diagnosis. Thus, the variables identified in the present study must be combined with clinical judgment and should most appropriately be used to guide empiric therapy rather than as a substitute for making a definitive diagnosis. Nevertheless, given the inherent delays in achieving a specific diagnosis (results of histopathology, culture), the findings of this study provide some data with which to base decisions about empiric therapy.
How do the findings of the present study compare with other studies of PNs in SOT recipients? Using the inclusion criteria as defined in the methods section, we found six published studies of PNs in SOT recipients (Table 8). Combining the results of these six studies, a total of 142 patients were analyzed and included recipients of liver, lung, heart and heart-lung transplants (but not kidney or kidney-pancreas). Approximately 30% of cases of PNs were non-infectious (range, 0–45%) which is similar to our results (44% of cases noninfectious). The mean onset of PNs in these studies was somewhat earlier (315 days) compared to 1060 days in the present study. Unfortunately, none of these prior studies used statistical analyses to assess the relationship of radiographic or clinical features to PN diagnosis. Thus, to our knowledge, the present study is the first to systematically analyze the association between clinical and radiographic features and the etiology of PNs in SOT recipients using multivariate statistical methods.
Several strengths and limitations of the present study should be noted. The strengths include the relatively broad range of variables analyzed, inclusion of all common organ transplant types, and statistical analyses used. Potential limitations include the retrospective study design and the moderate sample size. However, this represents the largest study of PNs in SOT recipients reported to date. We acknowledge that multiple testing was not specifically adjusted for in the statistical analyses and it is thus possible that some of the observed associations may have been due to chance alone. We recognize this is a potential limitation inherent in any study that involves multiple exploratory analyses. However, we were careful to include only variables that were biologically relevant and/or previously associated with specific etiologies of PNs in transplant recipients.
In conclusion, the present study has demonstrated that specific clinical and radiographic features are associated with specific etiologies of PNs in SOT recipients. These features, when combined with clinical judgment, might be helpful in guiding empiric therapy while results of definitive diagnostic procedures are pending. Future studies in other SOT settings to confirm these findings and to formally assess their utility for guiding empiric therapy decisions are warranted.
We gratefully acknowledge the expert secretarial assistance of Elaine Brooks.
This study was supported in part by the Fialkow Scholar Award (to A. P. Limaye) from the University of Washington Department of Medicine.