Our review population was orthotopic SOTRs with DTC. In accordance with published guidelines, we defined DTC as a cancer of donor origin in an SOTR, which was known or assumed to be present in the donor at the time of transplantation (8). Importantly, we excluded cases of DDC. Studies were eligible if they described recipients of liver, heart, lung, or intestinal transplants with DTC, and reported transplant type, transmitted cancer type, presentation or management, and patient survival (i.e., vital status at the time of reporting). We included any publication type except review articles and editorials.

We searched MEDLINE (1946 to present), EMBASE (1974 to present), PubMed (e-publications ahead of print only), Scopus, and Web of Science Core Collection. Our search terms included “cancer,” “tumour,” “transplant,” “donor,” “transmission,” and all related terms. We limited our search to human studies but did not apply date or language restrictions. We used publicly available search filters to restrict our search to cohort studies, case-control studies, case series and case reports, because we did not expect to find any interventional studies (19). We then searched “grey” literature sources including non-indexed conference proceedings, thesis repositories, and the World Health Organisation “NOTIFY” library (20). Lastly, we hand-searched reference lists of included articles. We executed our search on January 16, 2020. Our full search strategy is in the Supplementary Material.

Two reviewers (GG, MI) independently screened titles and abstracts followed by full-text review to determine study eligibility. We resolved disagreements by discussion. Where cancer origin was unclear (DTC vs. DDC) we involved a senior author (CW) or contacted authors for clarification. We cross-checked all included cases to identify duplicates between publications and included the report with the most complete information on each case.

Two reviewers (GG, MI) independently scored the quality of each included study using tools published by the Joanna Briggs Institute (JBI), with resolution of disagreements by discussion (21, 22). These are tools designed to assess the methodological quality of a study objectively, using categorical responses (yes/no/unclear/not applicable) to questions on key domains (e.g., “was the current clinical condition of the patient on presentation clearly described?”). We assessed case reports and registry studies against the JBI checklists for case reports and prevalence studies, respectively. We did not exclude any studies on the basis of quality. Because we did not find any reports with a comparator group, we were unable to assess the risk of bias.

Two reviewers (GG, UD) independently extracted data from included studies using a pre-piloted proforma (see Supplementary Material), creating a separate record for each case included in our review. Our five main domains of interest were: mode of presentation, time to diagnosis, tumour extent, treatment, and survival time.

We recorded the publication type, year, and total number of DTC cases (including heterotopic transplants) in each article. We considered reports of multiple transplants from a single donor as case reports. For each DTC case, we recorded recipient demographics, transplant type, mode of presentation (symptoms, graft dysfunction, surveillance, post-mortem), time to diagnosis, primary tumour site and histology, and cancer extent at diagnosis (confined to allograft/distant metastases). Donor variables were: demographics, history of cancer, and time from cancer diagnosis to donation. Information on management comprised cancer-specific treatment (e.g., chemotherapy, radiotherapy, loco-regional therapy, tumour resection), re-transplantation (and time from diagnosis), and modification of immunosuppression. We recorded all time intervals in days if less than 1 month and in whole months if more than 1 month.

Our outcomes were patient death, cause of death, cancer remission and cancer recurrence (and time since remission). Unless stated otherwise, we assumed that treatment procedures with curative intent (re-transplantation, resection) achieved remission. Where articles reported death only, we assumed that remission was not achieved. Where information on these outcomes was missing from case reports, we contacted study authors by email.

We first examined data completeness across our main domains of interest (mode of presentation, time to diagnosis, tumour extent, treatment, survival time). We then tabulated donor, recipient, and tumour-related characteristics of all included cases. After analysing all cases, we stratified our dataset, first by transplant type and then by cancer type. We did this because both domains are relevant to scenarios encountered in clinical practice. We grouped cancer type by the site of the primary tumour (e.g., lung), unless we found only one histological type in a particular site (e.g., melanoma). We did not group our data by study type because we analysed information at individual case level.

We determined the range, median and interquartile range (IQR) of the time from transplantation to DTC diagnosis. We then calculated the proportion of tumours with spread beyond the allograft at diagnosis. We compared tumour extent between transplant types using the chi-squared test. We tabulated the treatment modalities reported. Among cases that received a second allograft, we determined the median time from diagnosis to re-transplantation. We then calculated the proportion of cases that achieved cancer remission. Among these, we summarised the treatment modalities received, the proportion that recurred, and the proportion that died.

Our main outcome was all-cause mortality, calculated as the proportion of cases that died after a diagnosis of DTC. Since we only included cases where survival was reported, the denominator here was all cases (or all within a group). We used all-cause mortality because some treatment modalities (e.g., re-transplantation) confer substantial risk, so this is the most relevant patient-related outcome. The lack of comparator groups in each study precluded meta-analysis of treatment effects. Due to the size and heterogeneity of our study, multivariable analyses were not appropriate (23).

To analyse survival time, we restricted our dataset to cases with follow-up of at least 6 months, or to death. We assessed the heterogeneity of cases included in this analysis by summarising the range of follow-up time. We determined the median survival time following DTC diagnosis in all cases, then stratified by transplant and cancer type (for the commonest cancers).

Post hoc, we explored factors that may influence survival among cases with sufficient follow-up (6 months, or to death). Due to substantial variation in follow-up between studies, we censored this analysis at 3 years from DTC diagnosis. We examined the relationship between survival time and (1) transplant type (2), tumour extent at diagnosis, and (3) re-transplantation, using Kaplan-Meier curves and log-rank tests. To reduce confounding, we restricted our analysis of re-transplantation to cases with tumour confined to the allograft at diagnosis. We did this because patients with disseminated cancer would not ordinarily be considered suitable for re-transplantation, making them an inappropriate comparator group. Owing to data sparsity, it was not appropriate to test for an association between cancer type and survival time. Lastly, we tallied the number of cases of DTC in heterotopic SOTRs that received organs from the same donors as our included cases, and the proportion that died.

We performed study screening with Covidence software (Veritas Health Innovation, Australia), data extraction with EpiData v4.6 (EpiData Association, Denmark), and data analysis with Stata v15.1 (StataCorp, United States).

Our search retrieved 2,308 articles. After title and abstract screening, we assessed 223 full texts against our inclusion criteria. Fifty-eight articles (49 case reports, 9 registry studies) published between 1987 and 2019 were eligible for inclusion (Figure 1). Our review population comprised 73 cases of DTC in orthotopic SOTRs (52 from case reports and 21 from registry studies). These 73 cases originated from 69 donors and were reported from North America (n = 37), Europe (n = 33), Asia (n = 1), Australia (n = 1) and South America (n = 1). Supplementary Table S1 shows the characteristics of all included studies.

Overall, the quality of included articles was acceptable. However, there was substantial variation between studies and across quality domains. Among case reports, the domains with the lowest quality were the clinical condition of the patient at presentation and after treatment. The quality of registry studies was lower; most provided insufficient information on study size, case identification methods and sample coverage. Supplementary Figure S1, and Supplementary Figure S2 summarise study quality scoring against the JBI checklists.

Among the 73 cases in our study, data completeness varied across our five domains of interest (Supplementary Table S2). The proportion of cases with information in each domain was: tumour extent, 84% (61/73); time to diagnosis, 89% (65/73); presentation, 73% (50/73); treatment, 67% (49/73); survival time, 60% (44/73). We contacted the study authors of eight cases with incomplete outcome data; four replied with supplementary information which was added to the dataset for analysis.

Table 1 summarises the characteristics of included cases. There were 52 liver (including one liver-intestine-pancreas transplant), 10 heart, and 11 lung recipients (including one heart-lung transplant). Median (IQR) recipient age at diagnosis was 53 (41–60) years; 51% (37/73) were male. Median (IQR) donor age was 50 (39–62) years. In 29/73 (40%) cases, a cancer had been found in the donor. Six of these were diagnosed before donor assessment (between 4 months and 32 years prior to death), while 23 were discovered after organ implantation.

Table 2 summarises the types of transmitted malignancies included in our study. The commonest histological types were melanoma (n = 10) and choriocarcinoma (n = 7). Supplementary Table S3 shows the histology of all included cases. The most frequent mode of presentation was with symptoms, in 24/73 (33%) cases. Other methods of case detection were surveillance imaging (either routine or targeted because of transmission risk, n = 14), graft dysfunction (n = 4), tumour markers (elevated β-human chorionic gonadotropin in transmitted choriocarcinoma, n = 3), or retrieval or implantation biopsy (n = 5). Four cases were diagnosed at recipient post-mortem only.

Time from transplantation to DTC diagnosis ranged from 0 days to 6 years. In total, 48/73 (66%) cases were diagnosed within 1 year, and 60/73 (82%) within 2 years. Median (IQR) time to diagnosis was 8 (4–12) months; this was similar across transplant types (Table 1). The cancer types with the shortest time to diagnosis were choriocarcinoma [median (IQR) 1.5 (1 to 3) months] and sarcoma [2.5 (1 to 8) months; Table 2].

At the time of diagnosis, 29/73 (40%) tumours were confined to the allograft while 31/73 (42%) had disseminated. Twelve cases (eight heart, two liver, one heart-lung) had distant metastases only, with no tumour in the allograft. There was strong evidence of an association between tumour extent and transplant type; 27% (14/52), 100% (10/10) and 64% (7/11) of liver, heart, and lung recipients, respectively, had tumour dissemination at diagnosis (χ² = 15.2, p = 0.001; Table 1). It also varied between cancer types; all cases of choriocarcinoma had spread beyond the allograft at diagnosis, whereas all intestinal tumours and 6/7 neuroendocrine tumours (NETs) were confined to the allograft (Table 2).

Excluding palliative management, 43/73 (59%) cases included treatment details (Table 3). The commonest treatment was systemic chemotherapy; this was used in 20 cases and was the main treatment in 14. Seven cases underwent tumour resection and seven received loco-regional therapies, comprising radio/chemo-embolisation (n = 3), radiofrequency ablation (n = 2), brachytherapy (n = 1), and extracorporeal proton therapy (n = 1). There were six reports of altered immunosuppressive regimens, comprising a switch from calcineurin inhibitors to sirolimus (n = 4) or everolimus (n = 2).

Thirteen cases underwent re-transplantation. All of these were liver recipients that had no tumour dissemination at diagnosis, including one case pre-emptively re-transplanted after a post-mortem donor cancer diagnosis (DTC from the first donor subsequently recurred) (24). Re-transplantation was performed at a median (IQR) of 4 months (4 days–6 months) following DTC diagnosis. Treatments received prior to re-transplantation were loco-regional therapy (n = 3), tumour resection (n = 1), and chemotherapy (n = 1). Following re-transplantation, 3/13 (23%) tumours recurred between 2 weeks and 3 years later, and three patients died (Table 4).

In total, 19/73 (26%) cases achieved cancer remission following treatment. The main treatment modalities in these cases were: re-transplantation (n = 12; one case with recurrence 2 weeks after re-transplantation was not considered to have achieved remission), tumour resection (without subsequent re-transplantation, n = 5), loco-regional therapy alone (n = 1), and chemotherapy alone (n = 1). Of the 19 cases with cancer remission, six (33%) subsequently experienced a recurrence between 10 months and 3 years later, and five (26%) died (three of which had recurrent cancer).

In total, 55/73 cases (75%) died. This includes four cases diagnosed at post-mortem. Forty-seven deaths were due to cancer, three were due to other causes (sepsis, pneumonia, variceal bleed), and in five cases the cause of death was not evident. All-cause mortality was 69% (36/52), 80% (8/10), and 100% (11/11) in liver, heart, and lung recipients, respectively. Mortality by cancer type ranged from 50% in tumours of unknown origin to 100% in melanoma and central nervous system (CNS) tumours (Table 2).

There were 49 cases (36 liver, 5 heart, 8 lung) with follow-up of at least 6 months or to death. Among these, survival after DTC diagnosis ranged from 5 days to 13 years. Overall, 1-year survival was 39% (19/49). Overall median (IQR) survival was 7 (2–31) months and 9 (2–36), 6 (3–23) and 2 (1–5.5) months in liver, heart, and lung recipients, respectively. There was some evidence of an association between transplant type and survival time (log-rank χ² 8.3, p = 0.02; Figure 2), with the shortest survival in lung recipients. Survival time varied between the commonest cancer types. Median (IQR) survival was 2 (1–7) months in melanoma, 2 (1–2) months in CNS tumours, 9 (3–36) months in NETs and 26 (2–48) months in genitourinary tumours.

Median (IQR) survival was 16 (7–37) months in tumours confined to the allograft (n = 25) and 2 (1–9) months in disseminated cancers (n = 22). There was strong evidence of shorter survival in cases with tumour dissemination at diagnosis (log-rank χ² 9.9, p = 0.002; Supplementary Figure S3).

There were 26 cases (23 liver, 3 lung) without tumour dissemination at diagnosis and with sufficient follow-up for survival analysis. Among these, 13 (all liver) underwent re-transplantation, and 13 (10 liver, 3 lung) did not. All-cause mortality was 23% (3/13) in re-transplanted cases and 85% (11/13) in cases that were not re-transplanted. Median (IQR) survival was 36 (13–40) months and 7 (1–16) months in cases that did and did not undergo re-transplantation, respectively. There was strong evidence of longer survival in re-transplanted cases (log-rank χ² 9.3, p = 0.002; Figure 3).

There were 44 cases of DTC in recipients of heterotopic transplants from the donors of the cases included in our review (42 kidney, 1 pancreas, 1 kidney-pancreas); 21 (48%) of these died.