Clinicopathologic Features of Tissue Granulomas in Transplant Recipients: A Single Center Study in a Nontuberculosis Endemic Region Open Access

Institutional review board approval was obtained for conducting the present study. We performed retrospective electronic medical chart reviews of HSCTR and SOTR at Yale New Haven Hospital in New Haven, Connecticut for a 10-year period (2009–2019). Patients were screened for the presence of tissue granulomas, as noted in formal pathology reports from tissue biopsy specimens obtained pretransplant and posttransplant. We identified patients with granulomas by obtaining all pathology reports from each patient and using the keyword “granuloma” to identify reports that referenced granulomas. Only tissue specimens containing granulomas as reported by the reading pathologist were included. Specimens composed of cytology alone were not included. In addition, we performed a limited, independent slide review of select cases. For patients with tissue granulomas, further data were collected including the presence or absence of clinical symptoms at the time of granuloma discovery. Symptoms were classified as either systemic or localized/organ-specific. Systemic symptoms included fevers, chills, night sweats, weight loss, severe anorexia, or malaise. Organ-specific symptoms included gastrointestinal, lower respiratory tract, upper respiratory tract, dermatologic, genitourinary, musculoskeletal, and neurologic symptoms. Localized pain was also considered a symptom if it led to biopsy of the affected site. The indication for biopsy was recorded including whether it was prompted by presence of symptoms or to further investigate incidentally discovered radiographic or laboratory abnormalities or as part of routine disease surveillance in asymptomatic patients. Further data were obtained on patient characteristics, immunosuppression regimen at time of granuloma discovery, change in immunosuppression after granuloma discovery, time from transplantation to granuloma discovery, histopathology, and microbiologic diagnostics including Mycobacterium tuberculosis polymerase chain reaction (MTB PCR) and interferon gamma release assay. Charts were also reviewed for presence of epidemiologic MTB risk factors, including country of birth, history of international travel, human immunodeficiency virus (HIV) seropositivity, and history of homelessness or incarceration. The final clinical diagnosis for each case was determined based on available microbiologic, histopathologic, and laboratory data as well as clinical determination of the treatment teams. A limited, independent slide review was also performed by the authors. We recorded hospital readmissions and mortality due to granuloma-related disease at 1, 3, and 12 months after granuloma diagnosis for those discovered posttransplant. Patients who had granulomas discovered before transplantation were followed for 1 year posttransplant.

We used statistical tests to identify factors associated with the presence of infectious granulomas. X²  or Fisher's exact test was used for categoric variables, and independent samples t-test was used for continuous variables. All tests were 2-tailed with significance at P < .05. JMP Pro 15.0.0 software (SAS Institute, Cary, North Carolina) was used for all data analysis.

Our chart review revealed 2139 transplant recipients of whom 487 (22.8%) were HSCTR and 1652 (77.2%) were SOTR. Among the 1652 SOTR, 1062 (64.3%) were kidney transplant recipients, 486 (29.4%) were liver transplant recipients, and 104 (6.3%) were heart transplant recipients. Granulomas were identified in 56 (2.6%) of 2139 total patients. The average age of patients with granulomas was 53.4 years (range, 13–73), 29 of 56 (51.8%) patients were male, and 42 of 56 (75%) were white.

Of 56 total granulomas, 18 (32.1%) were found in HSCTR, whereas 38 (67.9%) were found in SOTR (Table 2). The overall prevalence of granulomas in HSCTR was 3.7% (18 of 487) compared with 2.3% (38 of 1652) in SOTR. Of 38 granulomas identified in SOTR, 26 (68.4%) were identified in liver transplant recipients (5.4% [26 of 486] of all liver transplant recipients), 11 (28.9%) were identified in kidney transplant recipients (1% [11 of 1062] of all kidney transplant recipients) and 1 (2.6%) was identified in a heart transplant recipient (1% [1 of 104] of all heart transplant recipients). Fourteen (25%) of the 56 total granulomas were identified on pathology specimens obtained before organ transplant and 42 (75.0%) were identified after transplantation. Among granulomas found posttransplant, the median time between transplant and granuloma discovery was 1.57 years (range, 0.15–25.7). The most common locations of the 56 total granulomas were the liver (n = 18, 32.1%), lung (n = 14, 25.0%,), skin (n = 8, 14.2%), and lymph node (n = 5, 8.9%), although granulomas were also infrequently identified on biopsies from bone marrow, colon, nasal cavity, small bowel, finger, paraspinal muscle, kidney, and spleen. There were no substantial histopathologic differences observed in infectious compared with noninfectious etiologies. Specifically, the vast majority of granulomas (n = 51, 91.1%) were nonnecrotizing. Of 5 necrotizing granulomas, 3 were of an infectious etiology and 2 were noninfectious.

Symptoms prompted biopsy in 27 of 56 (48.2%) patients, whereas in the remaining 29 cases (51.8%) biopsies were obtained for further evaluation of incidentally discovered imaging or laboratory abnormalities or as part of routine surveillance. Among the 27 symptomatic patients, systemic symptoms were observed in 12 patients (44.4%) and fever was present in 9 patients (33.3%), gastrointestinal symptoms in 9 patients (33.3%), lower respiratory tract symptoms in 9 patients (33.3%), upper respiratory tract symptoms in 4 patients (14.8%), and dermatologic symptoms in 5 patients (18.5%).

Of 56 total granulomas, 16 (28.6%) were determined to be infectious in etiology (Table 3, Figure 1). Of the 16 infectious granulomas, 12 were found in SOTR (31.6% of all 38 SOTR granulomas) and 4 were found in HSCTR (22.2% of all 18 HSCTR granulomas). Eleven of 16 (68.8%) were found in symptomatic patients, and 5 of 16 (31.2%) in asymptomatic patients (Figure 1). On univariate analysis, infectious granulomas were significantly associated with the presence of symptoms (40.7% [11 of 27] of symptomatic patients versus 17.2% [5 of 29] of asymptomatic patients, P = .05) and diagnosis in the posttransplant setting (7.1% [1 of 14] of pretransplant diagnoses versus 35.7% [15 of 42] of posttransplant diagnoses, P = .05). The most common infectious etiologies among the 16 infectious cases included bartonellosis (n = 3; 18.8%), CMV (n = 3; 18.8%), cryptococcosis (n = 2; 12.5%), and bacterial skin and soft tissue infection (n = 2; 12.5%). Bartonellosis was diagnosed via a combination of diagnostic techniques as well as expert opinion of consulting infectious disease physicians. In 2 cases, bartonellosis was diagnosed via positive Bartonella PCR on tissue as well as positive serologies, and in 1 of these cases, positive coccobacilli seen on Warthin Starry stain. In the third case, the diagnosis was made via a greater than 4-fold rise in Bartonella IgG titers from acute to convalescent serum, known exposure to cats, compatible clinical syndrome with improvement on antibiotic treatment, and expert opinion of consulting infectious disease specialists (Figure 2, A). CMV hepatitis was diagnosed in 2 cases via positive CMV tissue immunostaining in conjunction with high-grade CMV viremia (Figure 2, B and C). In the third case, CMV hepatitis was diagnosed based on a combination of markedly elevated CMV plasma viral load, compatible clinical syndrome, including elevated transaminases and pancytopenia with atypical lymphocytosis, complete normalization of serum transaminase levels after treatment with valganciclovir, and expert opinion of consulting infectious disease physicians. This patient was also notably high risk for CMV disease being a CMV donor positive, recipient negative liver transplant recipient who was less than 1 year from transplant and not on prophylactic antiviral agents. Histopathology was nonspecific in this case and revealed a patchy mononuclear infiltrate with mild lobular hepatitis and no significant endotheliitis or hepatic steatosis. Cryptococcal infections were diagnosed via a positive tissue culture in 1 case and a positive Grocott methenamine-silver stain and mucicarmine stain in the other case (Figure 2, D through F). Bacterial skin and soft tissue infections were diagnosed via positive bacterial tissue cultures.

The most common biopsy sites for infectious granulomas were lung (n = 6), liver (n = 5), and skin (n = 3) (Table 2, Figure 1). MTB PCR was ordered in 5 of 56 cases (8.9%) in tissue specimens obtained from liver (n = 2), colon (n = 1), lymph node (n = 1), and muscle (n = 1). Three of these were ordered on specimens from symptomatic patients. All 5 were negative. No patients in this cohort had documented MTB epidemiologic risk factors or positive interferon gamma release assay.

Tissue granulomas were noninfectious in 40 of 56 (71.4%) cases (Table 3, Figure 1, Figure 2, G and H). The most common noninfectious etiologies among the 40 noninfectious cases were T-cell–mediated rejection (n = 5; 12.5%), inflammatory bowel disease (n = 3; 7.5%), and sirolimus induced pneumonitis (n = 3; 7.5%). The etiologies of 12 of 40 (30%) were not established, but were presumed to be noninfectious based on a combination of lack of positive microbiologic studies, clinical assessment of treating physicians, lack of clinical symptoms, and clinical stability without antimicrobial therapy and/or with increases in immunosuppression. Microbiologic workup was limited in many of the cases without an established diagnosis and included bacterial tissue culture in 3 of 12 cases, fungal tissue culture in 1 case, and acid-fast tissue culture in 0 cases. Fungal and acid-fast stains were performed in 4 of these cases and MTB PCR was performed in 1.

Of 56 total cases, the majority of patients (n = 44; 78.6%) were immunosuppressed at the time of granuloma discovery, primarily because of standard posttransplant maintenance or chemotherapy for hematologic malignancy. The overall level of immunosuppression was increased in 20 of 56 (35.7%) patients after granuloma discovery, decreased in 8 of 56 (14.3%) patients, and unchanged in 28 of 56 (50%) patients (Table 4). There were no granuloma-related deaths at 1 and 3 months. At 12 months, 1 patient died from granulomatous cryptogenic organizing pneumonia that worsened with tapering of immunosuppression post-HSCT. There were no cases of granuloma-related hospital readmission. No granulomas discovered pretransplant were associated with new-onset symptomatic disease after transplantation and with increased immunosuppression.

Tissue granulomas were uncommon in a large transplant population with an overall prevalence similar to that of the general population (Table 1). Though susceptible to a greater array of infectious and noninfectious processes associated with formation of granulomas, transplant recipients may be unable to mount a vigorous immune response critical for immune cell recruitment that leads to granuloma formation.²¹  The diminished immune response may also account for the paucity of necrotizing granulomas in this study, even with infectious causes.

The etiologies of infectious granulomas were generally consistent with findings from previously published literature among immunocompetent hosts, though bartonellosis was notably common in our cohort, and symptomatic with fever in all 3 cases. An important distinction with other published reports is the absence of mycobacterial (MTB and nontuberculous mycobacteria) and dimorphic fungal causes in our cohort. However, no patients had epidemiologic risk factors for MTB and the study was performed in a single center in a geographically nonendemic area. Patients with both infectious and noninfectious granulomas had favorable outcomes. The lack of hospital readmissions and low-associated mortality at 1-year postgranuloma discovery, even with subsequent increases in immunosuppression in many patients, suggests that infectious diagnoses were not frequently missed.

Our data have meaningful clinical implications regarding the approach to newly discovered tissue granulomas in HSCTR and SOTR. The identification of granulomas in patients who have undergone transplantation should not reflexively trigger a barrage of microbiologic testing. Rather, patients should be risk stratified according to the likelihood of having an infectious etiology. Based on our study findings, the most important characteristics in determining the extent of microbiologic evaluation are the presence of clinical symptoms and epidemiologic risk factors for MTB and endemic fungi. Asymptomatic patients from non-MTB and nondimorphic fungal endemic areas in whom granulomas are discovered incidentally are unlikely to have an infectious etiology and thus in those patients, microbiologic testing may mostly be limited to standard tissue cultures and stains. Interestingly, of 5 cases with infectious etiologies in asymptomatic patients, 2 were associated with cryptococcal infection and 2 with CMV hepatitis suggesting that testing for these 2 pathogens may be considered. MTB PCR does not need to be routinely applied to tissue specimens in these patients. In contrast, patients in whom granulomas are discovered as part of an evaluation of clinical symptoms are at significantly higher risk of having an infectious cause. A wide variety of causative pathogens including bacterial, fungal, and viral causes were evident in this study. Given the limited number of infectious granulomas identified, it is not possible to make precise recommendations for specific testing. However, bartonellosis may be a more common etiology in transplant recipients and should be considered, particularly in patients with fever and other systemic symptoms.

In addition, this study suggests that granulomas discovered incidentally in asymptomatic patients before transplantation are unlikely to become clinically significant posttransplant. No patients in this study in whom granulomas were discovered before transplantation, including those without a known diagnosis, experienced new onset-symptomatic disease attributable to the granulomatous process after transplantation and with increased immunosuppression at 1 year posttransplant. Absent symptoms or a clear infectious diagnosis, routine posttransplant surveillance of incidentally discovered pretransplant granulomas is unlikely to be necessary.

There are several notable limitations to this study. Our findings are not generalizable to regions with higher prevalence of MTB and endemic fungi. In addition, a final clinical diagnosis was unable to be obtained in 12 of 40 (30%) noninfectious cases. These granulomas were presumed to be noninfectious based on the clinical judgment of the treatment teams, as well as the favorable outcomes seen without use of antimicrobial agents and increases in immunosuppression. However, without a confirmed clinical diagnosis it is possible that some of these granulomas may have been miscategorized. Of these 12 cases, only 3 had tissue cultures sent for bacteria, 1 had tissue cultures sent for fungi, and no tissue cultures were sent for acid-fast bacilli. Fungal and acid-fast stains were performed on 4 of these specimens. Still, it should be noted that the inability to confirm a diagnosis is consistent with prior literature in immunocompetent hosts in which 10% to 64% of granulomas are without an identifiable etiology.^4,7,9  Moreover, the positive outcomes among the patients in our study suggest that infections were not missed. In addition, it should be noted that 3 of 5 patients in whom liver allograft T-cell–mediated rejection was considered to be the etiology of the granulomatous process also had a medical history of other autoimmune diseases including primary sclerosing cholangitis, autoimmune hepatitis, primary biliary cholangitis, and inflammatory bowel disease (Table 5). Thus, it is possible that granulomas could have been secondary to these diseases and unrelated to allograft rejection, even if rejection was concurrently present. However, though it may not be possible to fully exclude these other causes, in all 5 cases, the clinical diagnosis of T-cell–mediated rejection was agreed upon by the treatment teams and was used to guide patient care. Similarly, 1 case of presumed CMV hepatitis was not confirmed histologically as the CMV immunostain on hepatic tissue was negative. While the patient did not have a clear alternate diagnosis, it is not possible to fully exclude other causes. However, the patient was a high-risk host for development of CMV disease, had a compatible clinical syndrome, and experienced marked improvement in serum transaminase levels after treatment with valganciclovir, suggesting that CMV hepatitis was the most likely diagnosis and was favored by consulting specialists. Last, the study is limited by a small sample size of 56 cases and further studies are needed to corroborate our findings. Moving forward, emerging diagnostics such as broad-range ribosomal PCR and metagenomic next-generation sequencing may help further elucidate the underlying etiologies of cryptogenic granulomas. In addition, given that an independent slide review was performed for only a limited number of specimens, there could be additional findings of relevance to the transplant population that future studies could address.

Tissue granulomas were uncommon in a large transplant population and mostly noninfectious. Patients with clinical symptoms at the time of granuloma discovery were significantly more likely to have an infectious cause compared with granulomas discovered incidentally, which were unlikely to be infectious. Bartonellosis and CMV hepatitis were the most frequently encountered infectious diagnoses, suggesting that these should be considered when evaluating for infectious etiologies. The approach to newly discovered granulomas in transplant recipients should involve risk stratification primarily based off of patient symptoms and epidemiologic risk factors. Extensive microbiologic testing, including use of MTB PCR may be safely avoided in low-risk patients and granulomas discovered incidentally before transplantation without a clear etiology likely do not require routine surveillance posttransplant.