Advertisement

Immunologic Research

, Volume 66, Issue 2, pp 219–223 | Cite as

T cell receptor-β J usage, in combination with particular HLA class II alleles, correlates with better cancer survival rates

  • Blake M. Callahan
  • Wei Lue Tong
  • George Blanck
Original Article

Abstract

T cell receptor (TCR) β V and J usage correlates with either the HLA class I or HLA class II major histocompatibility subtypes, and in both infectious diseases and autoimmune settings, the use of particular TCR-β V and J’s, in persons with specific HLA alleles, represents either better outcomes or certain clinical features. However, the relationship of TCR V and J usage, HLA alleles, and clinical parameters in the cancer setting has been less well studied. Here, we have evaluated the relationship of what is likely dominant TCR-β V and J usage among tissue-resident lymphocytes for lung, head and neck, kidney, stomach, ovarian, and endometrial cancers, with patient HLA class II alleles. The most striking indication is that TCR-β J subgroup usage, in combination with particular patient HLA class II alleles, correlated with either better or worse outcomes for lung cancer. One combination, TCR-β J2 segment usage and the HLA-DRB1*1501 allele, correlated with a better survival rate for both lung and head and neck cancers. These results fill a gap in knowledge regarding the relevance of HLA typing to cancer and indicate that HLA typing, along with an indication of dominant TCR-β J usage among tissue-resident lymphocytes, can be useful for prognosis.

Keywords

T cell receptor-β HLA class I and class II proteins Antigen presentation Cancer immune response The Cancer Genome Atlas 

Abbreviations

HNSC

Head and neck squamous carcinoma

KIRC

Kidney and renal carcinoma

LUAD

Lung adenocarcinoma

OV

Ovarian cancer

TCGA

The Cancer Genome Atlast

UCEC

Uterine corpus and endometrial carcinoma

Introduction

The advent of cancer has long been attributed to the failure of the immune system to detect non-self changes in cellular components, including viral oncogenes and mutant human genes. The primary mechanism by which the immune system adapts to intracellular and extracellular changes is through the binding of major histocompatibility complex (MHC) I and II proteins, in conjunction with T cell receptors (TCRs), to available non-self epitopes. Thus, it is reasonable that these interactions are critical to the development of cancer, and in fact, MHC class I and II proteins (HLA proteins in humans) have been linked to the susceptibilities of a variety of cancers, e.g., [1, 2].

The development of next-generation sequencing and computational advancements has made it possible to simultaneously detect both interacting components of the adaptive immune system (MHC class I and II proteins, and TCR V and J usage), for each patient, over very large cancer patient databases. In this report, we evaluated whether distinct survival patterns within cancer datasets could be detected by first identifying relatively common of TCR-β V and J segments among the TCR recombinations, and HLA class II alleles.

Methods

Recovery of TCR-β recombination reads from tumor specimen whole exome files

Whole exome (WXS) files were used to recover the TCR-β recombination, V and J segment reads representing tumor-resident lymphocytes using a previously described algorithm and protocols [3, 4, 5, 6]. Briefly, the WXS files were searched at a low stringency for all seven immune receptor V and J regions, and the recovered reads are then verified by automated interrogation of the Immune Gene Tics (IMGT) database. Multiple versions of the script for this purpose are in the supporting online material (SOM) of the indicated references. For this report, only the TCR-β reads were analyzed. In some cases, the reads representing the TCR-β V and J usage have been previously described [3, 4, 5, 6].

Determination of HLA class II alleles for The Cancer Genome Atlas

The HLA class II alleles for the HLA-DPB1, HLA-DQB1, and HLA-DRB1 were determined from (WXS) datasets in The Cancer Genome Atlas (TCGA) database. The relevant “slices” of the WXS BAM files were downloaded according to Genome Data Commons protocols [7], under the approval of dbGaP project no. 6300. The downloaded BAM file slices were evaluated using the xHLA software package [8], installed at USF Research Computing, from a source at GitHub.com.

Matching patient HLA class II alleles with TCR-β recombinations

Once obtained, the HLA class II alleles were matched to TCR-β recombination segments originating from the same barcode (patient). This was achieved by a simple original script termed VDJ-AlleleMatcher.pl and which is available upon request by email to the corresponding author. The sub-segments, e.g., J2-1, were stripped off by the script, i.e., TRBJ2-1 was considered as “TRBJ2.” All TRBV segments were included in the analyses; however, TRBV alleles were not differentiated. These combinations were then counted and tabulated for further statistical analysis including marrying identified barcodes with corresponding clinical data for survival analysis.

Overall survival and disease-free survival assessments for most frequently observed TCR-β VDJ segment and HLA class II allele combinations

Barcodes for patients representing various TCR-β V or J segment, HLA class II allele combinations were assessed for survival patterns using Kaplan-Meier (KM) curves with the cBioPortal.org web tool, as well as via independent, local analysis from downloaded clinical data and the IBM SPSS software.

Results

We first matched all TCR-β V and J usage, as determined by TCR-β V(D)J recombination reads recovered from exome files (4–6), with HLA class II types, as also indicated by the exome files (8). Very few of the matches represented a sufficient number of barcodes to conduct survival analyses. For all V or J segment, HLA class II matches numbering 20 barcodes and above, we first determined whether the barcodes with those matches represented survival distinctions, versus all remaining samples within the cancer dataset at issue. We then determined whether the either the V or J segment usage, regardless of HLA class II allele type, or the HLA class II allele, regardless of V or J segment usage, represented a survival distinction. A summary of the survival patterns detected for the TCGA HNSC, KIRC, LUAD, OV, STAD, and UCEC datasets is in Table 1.
Table 1

Survival patterns of TCGA barcode (patient) groups representing TCR-β V or J segment usage, in a recovered TCR-β recombination read, paired with specific HLA class II alleles. Italicized text indicates that both the TCR-β gene segment and the HLA class II allele are required for a statistically significant association with the indicated outcome

  

Number of barcodes used to compare with all remaining samples in the dataset

Significant, overall survival p value

Significant disease-free survival p value

Significant p value for TCR J usage alone

Significant p value for HLA allele alone

LUAD

TRBJ2 DRB1*15:01

58

0.016 (better)

0.046 (better)

No

No

LUAD

TRBJ2 DQB1*06:02

62

0.0065 (better)

0.016 (better)

No

No

LUAD

TRBJ1 DQB1*05:01

54

No

0.035 (worse)

No

No

LUAD

TRBJ2 DRB1*07:01

52

No

0.021(worse)

No

No

HNSC

TRBJ2 DRB1*15:01

35

0.046 (better)

0.028 (better)

No

No

HNSC

TRBJ1 DQB1*03:02

32

0.049 (better)

No

No &

No

STAD

TRBJ2 DPB1*02:01

43

0.029 (better)

No

No

No &

KIRC

TRBJ1 DRB1*07:01

20

No

0.019 (worse)

No

Yes

UCEC

TRBJ1 DPB1*04:01

40

No

0.021 (better)

Yes

No

OV

TRBJ1 DQB1*06:02

27

0.0096 (worse)

No

No

No &

OV

TRBJ1 DRB1*15:01

29

0.0077 (worse)

No

No

No &

&When removing the barcodes that represented the indicated J-HLA class II allele combination, was independently not significant

Analyses of the TCGA HNSC dataset revealed two distinct TCR-β J segment, HLA class II allele combinations significantly associated with survival patterns. Barcodes with the TRBJ2, DRB1*1501 combination were found to have significantly increased overall survival (OS) and disease-free survival (DFS) when compared to all remaining barcodes in the dataset (p value < 0.046 OS; p value < 0.028 DFS). Neither the TRBJ2 component alone nor the HLA class II allele alone had statistically significant, distinct survival patterns when compared to all remaining barcodes in the dataset. The TRBJ1, DQB1*0302 combination was also found to be statistically, significantly associated with OS. Neither the TRBJ1 nor the DQB1*0302 component of the combination was found to be (independently) statistically associated with survival differences.

Analyses of the TCGA STAD dataset yielded one significant TCR-β gene segment, HLA class II allele combination. The TRBJ2, DPB1*0201 combination was found to have significantly increased OS when compared to all remaining barcodes (p value < 0.029). Neither the TRBJ2 nor the DPB1*201 components of the combination were found to be independently, statistically, significantly associated with survival differences.

Four statistically significant TCR-β gene segment, HLA class II allele combinations were found for the TCGA LUAD dataset. The TRBJ2, DQB1*0602 and TRBJ2, DRB1*1501 combinations were found to have both increased OS and DFS (TRBJ2, DQB1*0602: p value < 0.0065 OS; p value < 0.016 DFS) (TRBJ2, DRB1*1501: p value < 0.016 OS; p value < 0.046 DFS). The barcodes representing the separate TCR-β gene segment, HLA class II components of these combinations had no statistically significant, independent survival distinctions, when barcodes representing these components were compared to all remaining barcodes. The KM analysis of the survival rate of the TRBJ2, DQB1*0602 combination is depicted in Fig. 1. Barcodes representing this combination had an average survival 20.22 months greater than all remaining barcodes in the LUAD dataset (Table 2). The TRBJ1, DQB1*0501 combination was found to have significantly increased DFS (p value < 0.035). The TRBJ2, DRB1*0701 combination had significantly decreased DFS (p value < 0.021).
Fig. 1

Survival pattern of TCGA LUAD TRBJ2-DQB1*0602 barcodes versus all remaining barcodes. Gray line indicates barcodes with TRBJ2-DQB1*0602. Black line indicates all remaining barcodes. The difference in the OS means was 20.33 months, where TRBJ2-DQB1*0602 had an increased survival rate compared to all remaining barcodes (log rank p < 0.006)

Table 2

Mean overall survival differences (in months) between barcodes with significant TCR-β VDJ-HLA class II combinations where each arm of the combination is independently insignificant in comparison to all remaining barcodes. Negative values indicate worse overall survival and positive values indicate better overall survival

Dataset

 

Difference between indicated TCR-HLA II combination and remaining barcodes for overall survival, average time (months)

LUAD

TRBJ2 DRB1*15:01

14.15

LUAD

TRBJ2 DQB1*06:02

20.33

HNSC

TRBJ2 DRB1*15:01

9.83

HNSC

TRBJ1 DQB1*03:02

NA

STAD

TRBJ2 DPB1*02:01

NA

OV

TRBJ1 DQB1*06:02

− 23.75

OV

TRBJ1 DRB1*15:01

− 24.893

TCGA KIRC revealed one significant TCR-β J segment, HLA class II allele combination: TRBJ1, DRB1*0701 had decreased DFS in comparison to the remaining barcodes in the dataset (p value < 0.019). However, in this case, barcodes representing the DRB*0701 allele also showed significantly decreased DFS.

The TCGA UCEC dataset included one significant TCR-β J segment, HLA class II allele combination, with the TRBJ1, DPB1*0401 combination representing increased DFS (p value < 0.021). Neither the TRBJ1 segment nor the HLA class II component of this association had an independently significant survival pattern.

The TCGA OV dataset was found to have two significant TCR-β J segment, HLA class II allele combinations for OS. Both TRBJ1, DQB1*0602 and TRBJ1, DRB1*1501 combinations demonstrated worse OS in comparison to all remaining barcodes. The average survival differences are shown in Table 2. Both of these combinations had independently insignificant TcR-β J segment, HLA class II allele components. The OS represented by the barcodes with the TRBJ1, DRB1*1501 combination is illustrated in Fig. 2.
Fig. 2

Survival pattern of TCGA OV TRBJ1-DRB1*1501 barcodes versus all remaining barcodes. Gray line indicates barcodes with TRBJ1-DRB1*15:01. Black line indicates all remaining barcodes. The difference in the OS means was 24.893 months, where TRBJ1-DRB1*1501 had a decreased survival rate compared to all remaining barcodes (log rank p < 0.0077)

Discussion

In the above approach, we found TCR-β J segment groups, in combination with particular HLA class II alleles, correlated with better or worse cancer survival rates. The results have several limitations. First, as with any correlative study, cause and effect relationships are not possible to establish. Further information regarding cause and effect may be inspired by the above work, for example, by using animal models where it would be possible to engineer the use of different combinations of TCR-β V and J usage, and MHC alleles. Second, it has not been possible to establish a replicative set. This is largely due to the impossibility of reassembling a set of patients with the same HLA class II allele distribution and treatment history, and due to the fact that random number generator subdivisions of the sets used in this report lead to samples sizes that are too small for a statistical analysis. However, these results are reproduced over several cancer types. And, the lack of association of survival rates with either the HLA class allele “arm” or the TRB J segment arm, in most of the cancer datasets above, supports the specificity of effect of the combination. Furthermore, as discussed in more detail below, important biological considerations provide context for these results, for examples, the known role of the T cell receptor, MHC interactions; the immunogenicity of cancer; and the role of the T cell receptor, HLA allele combinations in other diseases.

The results presented raise the question of whether the J segments bind better or worse to the HLA class II alleles where there is a distinctive survival rate. Interestingly, the J1 and J2 segments do have amino acid (AA) motifs exclusive to each of the two J segment groups. Three of the J1 segments have a 3′ LTVV AA motif in common, and four of the J2 segments have an LTVL motif in common. In both cases, these motifs do not appear in the other J segment set. However, it does need to be kept in mind that other members of the J1 and J2 segment groups have different motifs, and in some cases, these motifs are shared across the J1 and J2 groups. In addition to binding affinities, or lack of binding affinities, between TCR AA motifs and HLA AA motifs, other factors may play a role. For example, it is possible that certain J segments and certain HLA alleles are expressed at a higher level, and the high level of expression simply mediates a more robust anti-tumor response.

While there have been several reports regarding the combination of TCR gene segment usage and HLA alleles for infectious disease and autoimmunity settings [9, 10, 11, 12, 13, 14], with indications of such combinations correlating with distinct aspects of disease, there have been no such reports, particularly with survival rate connections, in the cancer setting. The data reported here provide an opportunity for enhancing the use of immunogenomics for prognostic purposes and indicate potential, relatively sophisticated biomarkers for attempts to understand which patients respond best to immunotherapies and why. In particular, the heavy use of immune checkpoint inhibitors has revealed only partial success. Are the best candidate for these therapies patients with tumor-infiltrating lymphocytes (TIL) where there is a known correlation of V(D)J usage, HLA allele, and disease progression? Likewise, interleukin-2 (IL-2) therapy for renal cell carcinoma has a relatively poor success rate, again leading the to the question, would there be greater success in using IL-2 for the treatment of patients who have TIL where a particular V(D)J usage is highly represented, and where that V(D)J usage occurs in persons with particular HLA alleles?

Notes

Acknowledgements

Authors would like to acknowledge the support of USF Research Computing and the taxpayers of the State of Florida. BMC was a recipient of a Bonati scholarship.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Supplementary material

12026_2018_8990_MOESM1_ESM.pdf (6 mb)
ESM 1 (PDF 6142 kb)

References

  1. 1.
    Qin N, Wang C, Zhu M, Lu Q, Ma Z, Huang M, et al. Fine-mapping the MHC region in Asian populations identified novel variants modifying susceptibility to lung cancer. Lung Cancer. 2017;112:169–75.  https://doi.org/10.1016/j.lungcan.2017.08.016.CrossRefPubMedGoogle Scholar
  2. 2.
    Gragnani L, Fognani E, De Re V, Libra M, Garozzo A, Caini P, et al. Notch4 and mhc class II polymorphisms are associated with hcv-related benign and malignant lymphoproliferative diseases. Oncotarget. 2017;8(42):71528–35.  https://doi.org/10.18632/oncotarget.17655.CrossRefPubMedPubMedCentralGoogle Scholar
  3. 3.
    Gill TR, Samy MD, Butler SN, Mauro JA, Sexton WJ, Blanck G. Detection of productively rearranged TcR-alpha V-J sequences in TCGA exome files: implications for tumor immunoscoring and recovery of antitumor T-cells. Cancer Inform. 2016;15:23–8.  https://doi.org/10.4137/CIN.S35784.CrossRefPubMedPubMedCentralGoogle Scholar
  4. 4.
    Samy MD, Tong WL, Yavorski JM, Sexton WJ, Blanck G. T cell receptor gene recombinations in human tumor specimen exome files: detection of T cell receptor-beta VDJ recombinations associates with a favorable oncologic outcome for bladder cancer. Cancer Immun Immunother. 2017;66(3):403–10.  https://doi.org/10.1007/s00262-016-1943-1.CrossRefGoogle Scholar
  5. 5.
    Tong WL, Tu YN, Samy MD, Sexton WJ, Blanck G. Identification of immunoglobulin V(D)J recombinations in solid tumor specimen exome files: evidence for high level B-cell infiltrates in breast cancer. Hum Vaccin Immunother. 2017;13(3):501–6.  https://doi.org/10.1080/21645515.2016.1246095.CrossRefPubMedPubMedCentralGoogle Scholar
  6. 6.
    Tu YN, Tong WL, Samy MD, Yavorski JM, Kim M, Blanck G. Assessing microenvironment immunogenicity using tumor specimen exomes: co-detection of TcR-alpha/beta V(D)J recombinations correlates with PD-1 expression. Int J Cancer J Int du cancer. 2017;140:2568–76.  https://doi.org/10.1002/ijc.30675.CrossRefGoogle Scholar
  7. 7.
    Grossman RL, Heath AP, Ferretti V, Varmus HE, Lowy DR, Kibbe WA, et al. Toward a shared vision for cancer genomic data. N Engl J Med. 2016;375(12):1109–12.  https://doi.org/10.1056/NEJMp1607591.CrossRefPubMedGoogle Scholar
  8. 8.
    Xie C, Yeo ZX, Wong M, Piper J, Long T, Kirkness EF, et al. Fast and accurate HLA typing from short-read next-generation sequence data with xHLA. Proc Natl Acad Sci U S A. 2017;114(30):8059–64.  https://doi.org/10.1073/pnas.1707945114.CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Di Sante G, Tolusso B, Fedele AL, Gremese E, Alivernini S, Nicolo C, et al. Collagen specific T-cell repertoire and HLA-DR alleles: biomarkers of active refractory rheumatoid arthritis. EBioMedicine. 2015;2(12):2037–45.  https://doi.org/10.1016/j.ebiom.2015.11.019.CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Weiskopf D, Angelo MA, de Azeredo EL, Sidney J, Greenbaum JA, Fernando AN, et al. Comprehensive analysis of dengue virus-specific responses supports an HLA-linked protective role for CD8+ T cells. Proc Natl Acad Sci U S A. 2013;110(22):E2046–53.  https://doi.org/10.1073/pnas.1305227110.CrossRefPubMedPubMedCentralGoogle Scholar
  11. 11.
    Miles JJ, Bulek AM, Cole DK, Gostick E, Schauenburg AJ, Dolton G, et al. Genetic and structural basis for selection of a ubiquitous T cell receptor deployed in Epstein-Barr virus infection. PLoS Pathog. 2010;6(11):e1001198.  https://doi.org/10.1371/journal.ppat.1001198.CrossRefPubMedPubMedCentralGoogle Scholar
  12. 12.
    Glanville J, Huang H, Nau A, Hatton O, Wagar LE, Rubelt F, et al. Identifying specificity groups in the T cell receptor repertoire. Nature. 2017;547(7661):94–8.  https://doi.org/10.1038/nature22976.CrossRefPubMedPubMedCentralGoogle Scholar
  13. 13.
    Quandt JA, Huh J, Baig M, Yao K, Ito N, Bryant M, et al. Myelin basic protein-specific TCR/HLA-DRB5*01:01 transgenic mice support the etiologic role of DRB5*01:01 in multiple sclerosis. J Immunol. 2012;189(6):2897–908.  https://doi.org/10.4049/jimmunol.1103087.CrossRefPubMedPubMedCentralGoogle Scholar
  14. 14.
    Klarenbeek PL, Doorenspleet ME, Esveldt RE, van Schaik BD, Lardy N, van Kampen AH, et al. Somatic variation of T-cell receptor genes strongly associate with HLA class restriction. PLoS One. 2015;10(10):e0140815.  https://doi.org/10.1371/journal.pone.0140815.CrossRefPubMedPubMedCentralGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of Molecular Medicine, Morsani College of MedicineUniversity of South FloridaTampaUSA
  2. 2.Immunology ProgramH. Lee Moffitt Cancer and Research InstituteTampaUSA

Personalised recommendations