Donor-derived follicular lymphoma after kidney transplantation — a case report

Introduction:

Solid organ transplant recipients have approximately three-fold increased risk of a subsequent cancer diagnosis compared to age-matched controls.¹ Several studies highlight the diverse histologic origins and frequent viral etiologies of these cancers, including post-transplant lymphoproliferative disorders (PTLDs) typically related to Epstein-Barr virus (EBV).² PTLD exists along a spectrum, with monomorphic PTLD comprising a heterogeneous set of clonal malignancies fulfilling the pathologic criteria for lymphoma in immunocompetent patients, most commonly diffuse large B cell lymphoma (DLBCL).³

Low-grade B cell neoplasms, such as follicular lymphoma (FL), arising post-transplant are not formally considered PTLD. FL is the most common indolent B cell lymphoma in non-transplant patients.⁴ The hallmark t(14;18) is found in ~85% of cases and produces an IGH::BCL2 rearrangement; it is also occasionally found in healthy individuals who do not develop FL,^5,6 highlighting the requirement for additional mutations to drive lymphomagenesis.^7,8

While donor-derived PTLD has been reported after allogeneic hematopoietic cell transplantation,⁹ donor-derived cancers of any histology after solid organ transplantation are exceedingly rare when donors have no prior cancer history.¹⁰ We present a surprising case of a father-daughter, kidney transplant donor-recipient pair both later diagnosed with FL (Figure 1). Genomic profiling confirmed the donor (i.e., father) origin of the recipient’s (i.e., daughter) FL and provides unique insight into FL pathogenesis.

Figure 1:

Timeline of the father’s and daughter’s clinical courses, highlighting the analyzed FL tumors (bold outlines) and PBMC germline normal samples. Created with BioRender.com

Case Presentation:

The patient is a woman with a history of amyloidosis originally diagnosed in 1999 at age 23 in the setting of extensive abdominal amyloid deposition incidentally noted during a cholecystectomy. In 2004, after evaluation of kidney failure at age 28 revealed renal amyloid deposition, she received a kidney transplant from her 48-year-old father. She received standard post-transplant immunosuppression with prednisone, mycophenolate mofetil, and tacrolimus.

In 2006, the father presented with supraclavicular lymphadenopathy. A lymph node biopsy showed effacement of the nodal architecture by follicles of lymphocytes expressing CD10, CD19, CD20, and CD22, leading to a diagnosis of FL. A CT scan revealed non-bulky lymphadenopathy in the cervical, supraclavicular, axillary, and inguinal regions, and splenomegaly. A bone marrow biopsy was negative for involvement by lymphoma, consistent with stage IIIA. He was asymptomatic with low tumor burden by GELF criteria and began observation. In 2009, a tonsillectomy showed involvement by FL. By 2010, progressive lymphadenopathy motivated treatment with six cycles of R-CHOP followed by maintenance rituximab for two years. Subsequently, a 2015 PET-CT scan to evaluate progressive cervical lymphadenopathy and back pain revealed new FDG-avid cervical lymph nodes and a paraspinal mass. He received two cycles of R-bendamustine before discontinuing due to fatigue and infections. Follow-up CT scans showed no evidence of disease until an isolated tonsillar recurrence in 2017 that has not required treatment.

Meanwhile, in 2016, the daughter presented at age 40 to an emergency room due to non-specific abdominal pain. A CT scan revealed modest retroperitoneal lymphadenopathy, and a lymph node biopsy showed a nodular arrangement of lymphocytes expressing CD10, CD20, and BCL2, consistent with FL. In-situ hybridization for EBER was negative. A PET-CT scan revealed FDG-avid paracaval, para-aortic, and aortocaval lymph nodes up to 2.0 × 1.0 cm, consistent with stage IIA. With a low tumor burden by GELF criteria but a potential relationship between her abdominal pain and retroperitoneal lymphadenopathy, she received four doses of weekly rituximab followed by maintenance rituximab for seven doses before discontinuing due to reported mouth sores. She remains disease-free.

Methods:

Informed consent was obtained from the daughter and father for tumor-normal genomic profiling and for publication of this study. Formalin-fixed, paraffin-embedded (FFPE) tissue from the daughter’s 2016 lymph node biopsy and the father’s 2009 tonsillectomy were obtained. Fluorescence in situ hybridization (FISH) was performed on the daughter’s biopsy using the Vysis CEPX SpectrumOrange and CEPY SpectrumGreen DNA Kit. FFPE tissue punches underwent DNA isolation via xylene extraction using the QIAamp DNA Micro Kit. Germline DNA was isolated from peripheral blood mononuclear cells (PBMCs) from the daughter and father using the QIAamp DNA Micro Kit. The father’s tumor and germline DNA, and the daughter’s germline DNA, underwent KAPA Hyper amplified library preparation and IDT Whole Exome capture. The daughter’s tumor DNA underwent SWIFT low-input whole genome library preparation. All libraries were sequenced on an Illumina NovaSeq 6000. Whole exome sequencing (WES) targeted 300X for the father’s tumor sample and 100X for the father’s and daughter’s germline samples, and whole genome sequencing (WGS) targeted 75X for the daughter’s tumor sample. FASTQ files were processed through an in-house pipeline¹¹ utilizing four variant callers (mutect,¹² strelka,¹³ varscan,¹⁴ pindel¹⁵). Variants exclusively within the exome capture space with allele depth >30 for tumor and normal samples across both patients were analyzed further, with manual review¹⁶ requiring five or more supporting reads across three fragments. Read density plots were generated using the karyoploteR package.¹⁷ Somatic hypermutation (SHM) motifs were quantified by a custom R script using the ‘attachContext’ function from the mutSignatures package.¹⁸

Results:

To interrogate the daughter’s FL for donor origin, we used two approaches to assess for Y chromosome DNA given the male sex of her kidney donor. By FISH, at least one copy of the Y chromosome was observed in 84.5% of the 200 analyzed nuclei from the daughter’s FL. Copy-number analysis of the daughter’s tumor-normal sequencing data revealed the presence of Y chromosome reads exclusively in the tumor sample (Figure 2A). These findings confirm the presence of male-derived tissue in the daughter’s FL biopsy.

Figure 2:

Evidence supporting a shared common precursor for the father’s and daughter’s lymphomas. (A) Density plots of reads mapping to X and Y chromosomes in the father’s and daughter’s PBMC normal and FL tumor samples. (B) Scatter plot of VAF by variant source for all high confidence mutations passing manual review. Non-silent or 5’ UTR mutations in COSMIC Tier 1 and 2 genes are labeled. (C) Phylogenetic tree showing the evolution of the father’s and daughter’s FL tumors. Branch length is proportional to the number of mutations unless crossed. Non-silent or 5’ UTR mutations in COSMIC Tier 1 and 2 genes are listed.

We next identified somatic variants in each of the two tumors. Initial analyses of the father’s tumor-normal pair revealed 233 variants, including alterations in key FL genes such as CREBBP, KMT2D, FOXO1, and STAT6.^7,8,19,20 For the daughter’s tumor sample, we performed a standard tumor-normal analysis, and additionally cross-referenced the called variants against the father’s PBMC germline sample. A total of 1486 preliminary variants were identified when comparing the daughter’s tumor to her paired PBMC germline sample, of which 1266 were present in the father’s PBMC germline sample, providing additional evidence of the donor origin of the daughter’s FL.

Given these findings and the potential for the daughter’s donor-derived FL to contain non-malignant cells of her own, we imposed additional stringent filtering to exclude alterations in either tumor sample identified in any germline sample with >1% allele frequency. The remaining variants were manually reviewed¹⁶ to obtain 240 high confidence somatic mutations: 76 unique to the daughter, 133 unique to the father, and 31 shared between both tumors (Figure 2B–C, Figure 3, Supplemental Table 1). Assessing the contribution of SHM to the mutational landscape of each tumor, we identified the canonical WRC/GYW motif in 25% (19/76) of the daughter-specific variants, 29% (39/133) of the father-specific variants, and 26% (8/31) of the shared variants (p > 0.05). Notably, of the 31 shared mutations, four impacted COSMIC Tier 1 and 2 genes²¹ suggesting potential pathogenic consequences: ATP1A1, CREBBP, KMT2D, and ZNFR3. Interestingly, both tumors possessed unique alterations in BTG1. Given reported frequencies of BTG1 mutations in FL of ~7–10%,^20,22,23 this observation is unlikely to be due to chance alone (p < 0.01); this shared pathway perturbance in these clonally related lymphomas may have pathobiological relevance.

Figure 3:

High confidence, non-silent or 5’ UTR shared and unique somatic mutations in COSMIC Tier 1 and 2 genes passing manual review in the father’s and daughter’s FL tumors.

Discussion:

To our knowledge, we report the first case of donor-derived FL after solid organ transplantation. The daughter recipient developed FL 12 years post-transplant, a decade after the father donor had developed FL, two years after donating his kidney. Using FISH and next-generation sequencing, we demonstrate that the daughter’s FL is donor-derived and shares multiple mutations with the father’s FL in key genes.

In a prior seminal study of 520,000 healthy participants, 100 subsequently developed FL, of which 56 (56%) had antecedent peripheral blood (PB) samples with detectable IGH::BCL2.⁶ However, 63 (29%) of 218 matched control individuals also had antecedent PB samples with detectable IGH::BCL2, usually at lower levels. The modest ~60% positive predictive value (PPV) for subsequent FL diagnosis with a permissive IGH::BCL2 threshold (>5×10⁻⁶) versus ~85% PPV with a stringent threshold (>1×10⁻⁴) suggests clonal expansion of FL precursors, perhaps via acquisition of cooperating mutations, is a feature of follicular lymphomagenesis.

Here, the mutational landscapes of the two related FL samples support several important concepts. First, the shared specific mutations in CREBBP and KMT2D align with prior observations of the typical temporal order of mutations in FL. Namely, CREBBP mutations followed by KMT2D mutations are most enriched within the earliest FL progenitors.^24,25 It follows that a premalignant population possessing these plus additional shared mutations was the common cell-of-origin from the father that gave rise to his FL and ‘contaminated’ his donated kidney to give rise to the daughter’s FL. Importantly, CREBBP has hotspot mutations in FL, such as the S1680 in-frame deletion observed in our unique case, which could occur independently in two unrelated lymphomas. However, the additional variants shared between the two lymphomas here, plus their informative sex chromosome data, provide confidence in the shared common precursor.

Further, the two versus 12 years between the kidney transplant procedure and development of FL in the father and daughter, respectively, highlight the indolent yet heterogeneous nature of FL. Interestingly, the father’s FL uniquely possessed mutations in several genes suspected to confer adverse risk in FL including FOXO1 and STAT6.^19,20 The father’s FL also possessed more total variants in putative driver genes than the daughter’s FL despite the shorter time for the father’s FL since the evolutionary branching at the transplant procedure. Speculating on possible explanations, the daughter’s post-transplant immunosuppression may have (1) permitted her disease to escape immune surveillance after acquiring fewer pathogenic variants, and/or (2) directly hindered mutation acquisition due to blunted TfH cell activity given the established role of cyclical germinal center re-entry in FL pathogenesis. It is also possible that our technical approach (75X WGS of the daughter’s limited lymph node core biopsy sample vs. 300X WES of the father’s more abundant excisional tonsillectomy sample) may have biased our analyses towards under-calling low VAF variants in the daughter’s tumor.

Finally, several examples exist of ‘favored pathways’ of lymphomagenesis, including paired diagnostic and relapsed DLBCL sharing patterns of mutated genes but not variants.²⁶ In our case, the two FL samples possess different BTG1 alterations. We hypothesize that the common FL precursor here may have been uniquely susceptible to the consequences of BTG1 dysregulation over other mechanisms. Overall, this unusual case provides a unique lens into the biology of FL.

Supplementary Material

PV_Data Supplement

Supplemental Table 1: All high confidence mutations in the father’s and daughter’s FL tumors passing manual review irrespective of impacted gene or mutation consequence.