Objectives: Pulmonary nodules are common in patients with hepatocellular carcinoma who are being evaluated for a possible liver transplant.
Materials and Methods: In this retrospective study, we analyzed the records of liver transplant recipients at our institution with a primary diagnosis of hepatocellular carcinoma who received transplants between 2000 and 2015. All patients had magnetic resonance imaging-confirmed disease within Milan criteria and a concurrent staging chest computed tomography. Patient survival was estimated using Kaplan-Meier methods and compared between pulmonary nodule characteristic groups. A Cox proportional hazards model was constructed for adjusted analysis.
Results: Of the 197 liver transplant recipients who met our study inclusion criteria (median follow-up, 40 mo), 115 (58.4%) had a total of 231 pulmonary nodules, with 57 (49.6%) having multiple nodules and 108 (93.9%) having nodules < 1 cm. The presence of pulmonary nodules did not negatively affect patient survival, per our univariate and multivariate analysis, nor did their presence affect their number, location, laterality, shape, edge, density, or the presence of calcifications (P > .05). However, pulmonary nodules > 1 cm were associated with decreased overall survival.
Conclusions: In our pretransplant evaluation of patients with hepatocellular carcinoma, pulmonary nodules < 1 cm did not portend worse patient or graft survival posttransplant.
Key words : Extrahepatic cancer, Metastases, Radiologic assessment, Transplant evaluation
A liver transplant (LT) can be curative in patients with hepatocellular carcinoma (HCC) whose disease burden adheres to the Milan criteria (1 tumor < 5 cm or < 3 tumors with the largest being < 3 cm, without vascular invasion or extrahepatic spread).1 Extrahepatic cancer is considered a contraindication to LT2,3; thus, its detection is critical in the pretransplant evaluation of patients with HCC.
According to autopsy studies, the lung is the most common site of extrahepatic HCC, with a reported incidence of 34% to 52%.4 However, more than 50% of patients with HCC have suspicious pulmonary nodules (PNs) visualized by computed tomography (CT).5 Therefore, before they are listed, LT candidates with PNs need a thorough pretransplant evaluation to rule out metastatic PNs.
Currently, no national guidelines govern the assessment of PNs in patients with HCC in the LT evaluation process. In patients with cirrhosis, a lung biopsy can pose significant risks, including death.6,7 Therefore, identifying the imaging characteristics of PNs that may be prone to malignancy is critically important. At our institution, we developed the following algorithm for LT listing: per the candidate's initial CT scan, PNs should be < 1 cm in diameter and stable, solitary, with rounded edges and calcifications. If PNs appear suspicious (ie, > 1 cm in diameter, multiple, changing in size, with spiculated edges and no calcifications), the candidate should undergo follow-up CT within 6 months; if the PN size is reported as unchanged, we list for LT.
In this retrospective study spanning 16 years, we analyzed the prevalence and, in particular, the clinical significance of PNs in patients with HCC who became LT recipients. Our goal was to identify PN characteristics that may prove helpful for distinguishing between malignant and benign lesions.
Materials and Methods
From January 2000 through December 2015, we performed 262 LTs in patients with HCC, diagnosed per the Milan criteria. During this time period, the mean time on the transplant wait list was approximately 12 months. For this retrospective study, our exclusion criteria included retransplants and the unavailability of outside CT scans for independent review. At our institution, we prospectively maintain all recipient information in a database approved by our institutional review board. In addition, we directly extracted multidisciplinary information, such as radiologic data, from the medical records. After collecting and analyzing all demographic, clinical, radiologic, and posttransplant data, we categorized our recipients according to the presence of PNs on their pretransplant chest CT scan.
All chest CT scans were performed at our institution with or without contrast, in helical multidetector acquisition through the chest, with a minimum of 16 channels. For review of the pulmonary parenchyma, images were reconstructed in axial and coronal 5-mm soft-tissue configurations, in 1-mm and 5-mm sharp lung window cuts, and with a maximum intensity projection of 20 mm.
Initial chest CT scans were reviewed for the diameter, number, location, laterality, shape, edge, and density of PNs, as well as for the presence of calcifications, by a blinded board-certified chest-dedicated radiologist (DCO) with 12 years of dedicated chest CT experience, using a Philips IntelliSpace PACS workstation (Amsterdam, the Netherlands). The largest dimension of the PNs for a given LT recipient was recorded, as well as their shape, location, edge, density, and associated calcifications. Per our programmatic algorithm, if PNs appeared suspicious (ie, > 1 cm in diameter, multiple, changing in size, with spiculated edges and no calcifications), a follow-up CT scan was performed within 6 months to evaluate any changes in the indexed PNs, any new PNs, the progression of the hepatic tumor burden, and new sites of metastasis. Pulmonary nodule diameter was classified as increased, decreased, or stable. For candidates who underwent follow-up CT, the mean imaging surveillance period was 4.3 months.
To summarize and compare demographic characteristics by the presence of PN, we used the chi-square test for categorical data and the Wilcoxon rank-sum test for continuous data. The primary outcome was time to death (measured in months from the date of LT to the date of death). To calculate the median survival estimates for each PN group, we used the Kaplan-Meier method; to compare the median survival time between the groups, we used the standard log-rank test.
For our adjusted analysis of time to death, we used a Cox proportional hazards model, with malignancy as a secondary outcome and with death from other comorbidities censored. We estimated the survival outcomes for each PN group, according to the largest nodule for each recipient, and compared outcomes between the groups.
Of the 262 LTs performed at our institution in patients with HCC during our study period (January 2000 through December 2015), 67 were excluded from our analysis: 12 because they were retransplants and 53 because of the unavailability of outside CT scans for our independent review. Thus, 197 LT recipients met our study's inclusion criteria (Figure 1). Of those 197 recipients, 82 (41.6%) did not have PNs detected during their pretransplant evaluation and were listed immediately. Of all the underlying causes of cirrhosis in our study cohort, hepatitis C infection overwhelmingly predominated (Table 1).
The remaining 115 patients (58.4%) had PNs detected on their initial CT scan; 86 of them were still listed immediately because of the benign appearance of their PNs (ie, < 1 cm, solitary, stable in diameter, with rounded edges and calcifications). Of those 86 recipients, 73 later underwent CT posttransplant; therefore, we had scans available to evaluate any progression of PN growth.
However, the other 29 patients (of our 115 LT recipients with PNs) had suspicious-appearing PNs (> 1 cm, multiple, changing in size, with spiculated edges and no calcifications). These 29 patients were thus not listed until after a follow-up CT scan showed no further change in the appearance of their PNs. Of these 29 recipients, 18 had a follow-up CT scan available for our independent review.
The mean posttransplant follow-up time in our study was 46 months (interquartile range [IQR], 15-73 mo). We found no statistically significant differences between recipients with or without PNs in terms of sex, race/ethnicity, or age (Table 2). Those with PNs tended to be somewhat older (mean age, 59) than those without PNs (mean age, 56), but the difference was not significant (P = .07).
Radiologic characteristics of pulmonary nodules
Of the recipients with PNs, 50% had multiple PNs (Table 3). Assigning the largest PN for each recipient, we found that most of the largest PNs (67%) were < 0.5 cm in diameter; 94% were < 1 cm. Most of the largest PNs were on the right side and the upper lobes. The overwhelming majority of the largest PNs had no calcifications. Most of the largest PNs were round or oval, with smooth edges and a solid density.
The presence of PNs did not negatively affect patient survival, per our univariate analysis. The median time to death in the group of recipients with PNs was 127 months (IQR, 45 to not reached); in the group without PNs, median time to death was 80 months (IQR, 20 to not reached; P = .06). For all 197 recipients, the unadjusted hazard ratio for death was 0.64 (95% confidence interval, 0.40-1.02); adjusted for age and time from the CT scan to the LT, the hazard ratio for death was similar: 0.65 (95% confidence interval, 0.40-1.04). Of the 69 recipients who died during our study period, 13 (38%) in the group with PNs died of malignancy-related causes compared with 12 (34%) in the group without PNs (P = .73).
In the group with PNs, the following variables were not associated with decreased survival (P > .05) per our multivariable analysis: the number of PNs and, according to each recipient's largest PN, the laterality, shape, density, edge, and presence of calcifications. However, for each recipient's largest PN, a diameter > 1 cm was associated with decreased overall survival (P = .02) (Figure 2). Of the 29 recipients with PNs > 1 cm, 6 died (Table 4). The 2 malignancy-related deaths in this group occurred at 94.5 and 86.1 months posttransplant.
In the group of recipients with suspicious PNs, 18 (62.1%) of the 29 had a follow-up CT scan available for our independent review. The median time between their initial and follow-up CT scan was 5.3 months (IQR, 4.2-9.5 mo). Of these 18 recipients, 1 (6%) had a follow-up CT scan that we could not evaluate because of its poor quality. For the other 17 patients (94%), we found that their largest PN was either stable or had decreased in diameter.
Of the 86 recipients with PNs who were nonetheless listed immediately because of the benign appearance of their PNs during their pretransplant evaluation, 73 later underwent chest CT posttransplant for a variety of unrelated diagnostic purposes (median time between their initial and follow-up CT scans, 32.7 months; IQR, 14.7-61.6 mo). Of note, these patients did not have a follow-up CT scan in the pretransplant phase for the purposes of surveying their PN; rather, the patients had a chest CT for an unrelated posttransplant clinical indication, which permitted the review of the images for the purposes of this study. In 71/73 recipients (97%), we found that their largest PN was either stable or had decreased in diameter; however, in 2 patients (2%), we noted an increase in diameter.
Pulmonary nodules are common in patients with HCC who are being evaluated for LT. Our findings were similar to those of previous studies: more than 50% of the LT recipients in our study had PNs at the time of their initial pretransplant evaluation.8,9 We found that LT recipients with and without PNs had similar survival rates, and a similar percentage in each group died as a result of malignancy. Lesions < 1 cm in diameter that otherwise met our institution's criteria (ie, solitary, stable in diameter over time, with rounded edges and calcifications) were likely to be clinically benign.
We did note that recipients with PNs > 1 cm had decreased posttransplant survival; however, because of this finding's high susceptibility to selection bias, its significance is questionable. That said, we believe that LT candidates with PNs > 1 cm should undergo a rigorous pretransplant evaluation before being listed, including histologic assessment of their PNs (or at least follow-up imaging), to rule out extrahepatic HCC or a de novo pulmonary malignancy, both of which are contraindications to LT listing.2,3
Interestingly, the 2 recipients in our study with PNs > 1 cm who died of HCC recurrence had late recurrence (94.5 and 86.1 mo posttransplant) and only in the liver. Furthermore, both of these recipients had HCC macrovascular invasion in the resected specimen, a finding that is consistent with previous studies. For instance, in a recent analysis that examined predictive factors for recurrence of HCC after LT (n = 364), Andreou and associates reported a recurrence rate of 25% at a median of 19 months posttransplant; in their series, the only predictive factor for HCC recurrence was macrovascular invasion in the resected liver.10 That predictive factor has also been demonstrated in similar studies.11-13 Worthy of mention is that, in our study, none of the 6 recipients with PNs > 1 cm who died had lung-specific disease. This was perhaps a phenomenon of selection bias, simply a coincidental finding, or possibly an indication of the biology of the disease.
The criteria that we used for patient evaluation and selection in our study are supported by other studies. Lee and colleagues, in their series of 172 LT recipients with HCC who had ≥ 1 PN, reported that 1.7% of them had HCC metastases and 1.7% had bronchopulmonary malignancy9; however, none of the malignant PNs that they discovered had a triangular or lentiform shape or had calcifications.
In a similar study, Concejero and associates evaluated 152 living-donor adult LT recipients in Taiwan to define the incidence and cause of PNs and to determine their impact on overall survival.8 They reported that only 9 study patients (5.9%) had PNs on their pretransplant chest CT scan, a rate that is substantially lower than they previously reported for deceased-donor LT recipients (as well as substantially lower than in our study). Perhaps that particular population of living-donor LT recipients in their study was somehow fundamentally different from deceased-donor LT recipients. All of the study patients had hepatitis B virus-related cirrhosis with HCC, with 2 undergoing thoracoscopic biopsy evaluation, revealing cryptococcosis, and 2 diagnosed with tuberculosis. None of these patients had pulmonary malignancy.
Our transplant center is in the Midwest region of the United States, where fungal infections such as histoplasmosis,14,15 aspergillosis,15 North American blastomycosis,16 and mucormycosis17 are endemic. One of the largest studies to date examining the malignancy potential of incidentally discovered PNs is the Mayo Lung Project, which enrolled 10 993 men (1971 through 1983) with a history of smoking who were not suspected to have lung cancer.18,19 As a continuation of that effort, the Mayo Clinic CT screening study (1999 through 2002) enrolled 1520 men from the original Mayo Lung Project who were > 50 years old and either current or former smokers of > 20 pack-years (if former smokers, they had to have quit within the prior 10 years). Each patient in the Mayo Clinic CT screening study had a baseline chest CT scan, followed by 3 subsequent annual CT scans, in addition to annual sputum and blood samples for cytologic analysis. In that study, the PN management recommendations were as follows: for PNs < 4 mm in diameter, a repeat CT scan in 6 months; for PNs of 4-8 mm, a repeat scan in 3 months; for PNs of 8-20 mm, a thin-slice CT scan, with positron emission tomography considered; and for PNs > 20 mm, a biopsy. Pulmonary nodules that remained stable in diameter for > 2 years were defined as benign. After 3 years of screening, the study identified 2832 uncalcified PNs. Ultimately, 40 primary lung cancers were diagnosed (1.4% yield), with 26 identified with baseline chest CT scan and 2 diagnosed by sputum cytology alone. Of those 40 study participants with primary lung cancer, 31 (77.5%) underwent pulmonary resection, with 21/31 patients (60%) having stage IA cancer.20
In our study, we aimed to offer transplant teams an algorithm of PN characteristics for HCC patients being evaluated for a possible LT. The Mayo Clinic experience is an ideal comparator and provides an excellent background for our study. Similar to our population, most of the patients in the Mayo Clinic series reside in the Midwest, where environmental and occupational hazards must be taken into consideration (in particular, histoplasmosis, aspergillosis, North American blastomycosis, and mucormycosis). Importantly, of the PNs identified in otherwise asymptomatic patients in the Mayo Clinic series, more than 98.5% were found to be benign. This is a key consideration in the evaluation of LT candidates within our geographic location.
Of note, differentiating active or prior fungal infections from malignancy can be challenging. In fact, Guimarães and associates recently reported that most malignancy-suspicious CT lesions ultimately end up being fungal in nature.21 Such lesions frequently have density, spiculation, and shape characteristics that closely mimic malignant-appearing PNs, making them indistinguishable from thoracic malignancies.21 Therefore, we advocate listing patients with HCC who have incidental asymptomatic PNs < 1 cm in diameter. For PNs > 1 cm, we advocate further evaluation, including histologic assessment (or at least follow-up imaging), to rule out extrahepatic HCC or a de novo pulmonary malignancy. In addition, we advocate obtaining a repeat lung-dedicated CT scan within 3 months because malignant pulmonary tumors can increase in diameter within that time interval.9
Our study has several limitations. First, as a single-center retrospective study, it is subject to the drawbacks of retroactive analyses of already acquired data at one institution. However, our study spanned 16 years and studied hundreds of LT recipients, all treated in a relatively uniform manner versus that shown with multicenter studies. Second, we excluded patients who did not qualify for LT listing (for medical or social reasons), as well as patients who were initially listed but then were delisted (for various reasons). Thus, some patients may not have been listed because of the discovery, during their pretransplant evaluation or workup, of a new or progressing malignancy. Third, the PNs in our study were not histologically assessed, other than a few PNs in autopsies performed on LT recipients who died posttransplant after their HCC progressed. Fourth, in our study, patients with PNs on their initial chest CT scan waited about 6 months before their LT. Therefore, it is possible that our recipients with PNs were healthier than those who did not survive for 6 months pretransplant, who were excluded from our study. The median time from the initial chest CT scan to LT for our recipients with PNs was 7.7 months (IQR, 4.2-15.4 mo); without PNs, it was 6.0 months (IQR, 2.8-13.5 mo). Our sample population does not represent all patients with HCC in need of LT, but our study showed that LT recipients with PNs at our institution generally have positive outcomes. This finding supports our policy of waiting 6 months to document the growth characteristics of suspicious PNs.
Volume : 16
Issue : 3
Pages : 314 - 320
DOI : 10.6002/ect.2017.0227
From the 1Division of Transplantation, Department of Surgery, and
the 2Department of Radiology, University of Minnesota Medical School,
Minneapolis, Minnesota, USA; the 3Department of Radiology, Cerrahpasa
Medical Faculty, University of Istanbul, Istanbul, Turkey; and the 4Biostatistics
and Bioinformatics Core, Masonic Cancer Center, and the 5Department
of Medicine, University of Minnesota Medical School, Minneapolis, Minnesota, USA
Acknowledgements: The authors have no sources of funding for this study and have no conflicts of interest to declare. The authors thank Mary Knatterud, PhD, for editorial assistance with this article.
Corresponding author: Oscar K. Serrano, Department of Surgery, Division of Transplantation, Mayo Mail Code 195, 420 Delaware Street SE, Minneapolis, MN 55455, USA
Phone: +1 612 625 5151
Table 1. Underlying Causes of Cirrhosis (N = 197)
Table 2. Recipient Characteristics
Table 3. Radiologic Characteristics of Pulmonary Nodules
Table 4. Cause of the 6 Deaths in Recipients with Pulmonary Nodules > 1 cm (n = 29)
Figure 1. Schematic Diagram of Recipient Selection
Figure 2. Patient Survival by Pulmonary Nodule Status