Intratumor heterogeneity as a predictor of disease progression in DLBCL

On 10 April 2019, Yue Wang from the University of Shanghai, Shanghai, CN, and colleagues, published in Carcinogenesis whole-exome sequencing data to assess intratumor heterogeneity in patients with diffuse large B-cell lymphoma (DLBCL).

Tumor heterogeneity plays a crucial role in tumor development, recurrence and metastasis and impacts the clinical diagnosis, as patients with high tumor heterogeneity have poor prognosis and poor therapeutic responses. In this study, the authors used a novel quantitative technique, mutant-allele tumor heterogeneity (MATH), through whole-exome sequencing data to evaluate the effects of intratumor heterogeneity (ITH) in the risk of progression in early stage DLBCL.

Study design & baseline characteristics

Discovery dataset (whole-exome sequencing data from the TCGA database):

• N = 22 patients with early stage DLBCL (Stage I and II)
• Median patient age (range): 51 (23−82) years
• Male patients: 40.9% (n = 9)
• Median follow-up: 25.6 months:
  • Four patients relapsed or progressed during follow-up
• Received primary treatment for DLBCL:
  • Rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP)-like chemotherapy: n = 19 patients
  • Rituximab, etoposide, prednisone, vincristine, cyclophosphamide and doxorubicin (R-EPOCH): n = 2 patients
  • CHOP: n = 1 patient
• Patient International Prognostic Index (IPI):
  • IPI ≤ 2: n = 21 patients
• Patient data were excluded:
  • Without available information from whole-exome sequencing and ‘Mutation Annotation Format (maf)’ files
  • Without information regarding age, sex, disease stage, histology type and follow-up time
  • If from patients with Stage III or IV DLBCL

Validation dataset (genomic data from The Gene Index Project database):

• N = 35 patients with early stage DLBCL (Stage I and II)
• All patients < 60 years old
• Male patients: 65.7% (n = 23)
• Median follow-up: 83.1 months:
  • Ten patients progressed or relapsed during follow-up
• Received primary treatment for DLBCL:
  • R-CHOP-like chemotherapy: n = 35 patients
• Patient IPI:
  • IPI ≤ 2: n = 35 patients
• Patient data were excluded:
  • Without information on genetic mutations
  • Without information regarding age, sex, disease stage, histology type and follow-up time
  • If from patients with Stage III or IV DLBCL

Key findings

Discovery cohort:

• Median MATH score: 23.9
• The cohort was divided into low and high MATH score groups according to median expression level:
  • High MATH group (n = 11): median, 30.0 MATH score
  • Low MATH group (n = 11): median, 20.8 MATH score
• Higher MATH score was associated with a higher risk of disease progression (PD; P = 0.045)

Validation cohort:

• Median MATH score: 45.3
• The cohort was divided into low and high MATH score groups according to median expression level:
  • High MATH group (n = 17)
  • Low MATH group (n = 18):
• No patients progressed in the low MATH group
• In the high MATH group, higher MATH score was associated with a higher risk of PD (P = 0.025)

Total cohort:

• Most popular somatic mutations found in DLBCL patients:
  • Immunoglobulin lambda variable 3 (IGLV3)
  • B-cell translocation gene 2 (BTG2)
  • Caspase recruitment domain-containing protein 11 (CARD11)
  • Immunoglobulin heavy variable 2 (IGHV2)
  • Immunoglobulin lambda constant 2 (IGLC2)
  • LDL receptor-related protein 1B (LPR1B)
  • P2Y receptor family member 8 (P2RY8)
  • Immunoglobulin heavy constant gamma 1 (IGHG1)
  • Ig heavy chain mu (IGHMM)
  • Suppressor of cytokine signaling 1 (SOCS1)
• IGLC2 was only found in the high MATH group (P = 0.016) while IGHG1 was only found in the low MATH group (P = 0.038)
• Multifactor Cox analysis showed that none of the following factors affected MATH expression:
  • Gender
  • Age
  • Race
  • Extranodal involvement
  • Tumor max dimension
  • Body mass index
• The authors also identified genetic mutations that were commonly observed in patients with early DLBCL and that promote cancer progression:
  • BTG2
  • CARD11
  • PIM1
  • P2RY8
• MATH score and molecular DLBCL subtypes:
  • Patients in the discovery set were classified by molecular subtype:
    • Germinal center B-like (GCB): 40.9% (n = 9)
    • Activated B-cell (ABC): 36.4% (n = 8)
    • Type 3: 22.7% (n =5)
  • No significant correlation was found between MATH score and DLBCL molecular subtype (P = 0.211, Student’s t-test; Spearman’s correlation test, P = 0.312)

Conclusions

• Higher ITH as measured by MATH was associated with a higher risk of DLBCL progression
• MATH score based on whole-exome sequencing was an independent risk factor in patients with early stage DLBCL, and could potentially be used for deciding clinical treatment in these patients