Liquid biopsy in extranodal NK/T-cell lymphoma: a prospective analysis of cell-free DNA genotyping and monitoring

Abstract

Satisfactory tumor material is often hard to obtain for molecular analysis in extranodal natural killer (NK)/T-cell lymphoma (NKTCL) at present. However, the accuracy and utility of circulating cell-free DNA (cfDNA) genotyping have not been adequately assessed in NKTCL. We therefore performed targeted next-generation sequencing on tumor tissues and a series of longitudinal plasma samples prospectively collected from a cohort of high-risk NKTCL patients. Concordance of genotyping results of paired baseline tumor and cfDNA and the predictive value of dynamic cfDNA monitoring were evaluated. At baseline, 59 somatic variants in 31 genes were identified in tumor and/or plasma cfDNA among 19 out of 24 high-risk NKTCL patients (79.2%). Plasma cfDNA had a sensitivity of 72.4% for detection of somatic variants identified in tumor biopsies before treatment. Plasma cfDNA also allowed the identification of mutations that were undetectable in tumor biopsies. These results were also verified in a validation cohort of an additional 23 high-risk NKTCL patients. Furthermore, longitudinal analysis showed that patients with rapid clearance of NKTCL-related mutations from plasma had higher complete remission rates (80.0% vs 0%; P = .004) and more favorable survival (1-year progression-free survival [PFS] rate, 79.0% vs 20.0%; P = .002) compared with those with persisting or emerging mutations in plasma. In addition, low cfDNA concentration before treatment was associated with favorable survival outcome for patients with NKTCL (1-year PFS, 90.0% vs 36.4%; P = .012). In conclusion, cfDNA mirrors tumor biopsy for detection of genetic alterations in NKTCL and noninvasive dynamic plasma cfDNA monitoring might be a promising approach for tracking response and survival outcome for patients with NKTCL.

Graphical abstract

Figure 1.

NGS of plasma cfDNA and tumor gDNA in NKTCL of the training cohort. (A) Overview of baseline mutational profiles in the entire training cohort. (B) Case-level genetic alterations of tumor gDNA and plasma cfDNA in 16 patients with paired plasma/tumor samples. Match indicates variants identified in both tumor gDNA and plasma cfDNA. Formalin-fixed paraffin-embedded (FFP) specific indicates variants identified in tumor gDNA only. Plasma (PLA) specific indicates variants identified in plasma cfDNA only. CN, copy number.

Figure 2.

Concordance between plasma cfDNA and tumor gDNA genotyping in 16 patients with paired tumor/plasma patients in the training cohort. (A) The percentage of tumor biopsy–confirmed mutations that were detected in cfDNA. (B) The number of mutations discovered in baseline plasma cfDNA and/or tumor gDNA. (C) Comparison of AFs of mutations identified in baseline tumor gDNA and plasma cfDNA. (D) Scatter plot of mutation AF in cfDNA vs AF in tumor gDNA for each variant.

Figure 3.

Longitudinal assessment of mutation abundance in cfDNA upon treatment. (A) NKTCL-related mutations disappeared during cfDNA monitoring among patients who achieved CR at the end of treatment. (B) NKTCL-related mutations persisted during cfDNA monitoring among patients who did not achieve CR at the end of treatment.

Figure 4.

Noninvasive real-time monitoring of cfDNA genotyping and disease changes in patients with NKTCL. Patient identification (ID) numbers 2 (A), 5 (B), 10 (C), 12 (D), 15 (E), 17 (F), 24 (G), 20 (H), 3 (I), 6 (J), 4 (K), 1 (L), and 11 (M). PR, partial remission.

Figure 5.

Survival curves of NKTCL patients included in this study. PFS (A) and OS (B) of the entire cohort. (C) Comparison of PFS between patients with a clearance of ctDNA and those with persisting detectable ctDNA during monitoring. (D) Comparison of PFS between patients with high and low baseline plasma cfDNA concentration.