Calibration and performance of Ig-HTS. (A) The overall success rate of tumor clonotype identification is shown per patient (left bar) and further broken down when considering number of biopsy specimens (right two bars). (B) When considering all samples assayed, fresh/frozen tumor biopsy specimens had a significantly higher rate of tumor clonotype detection compared with formalin-fixed paraffin-embedded (FFPE) specimens (Fisher exact, P = .007). (C) The success rate of tumor clonotype identification from biopsy specimens was associated with the available yield of DNA input for the reaction. An increasing success rate can be seen with increased input mass. Samples with less than minimum required DNA input (<1.5 ng) were not assayed and are therefore excluded from all panels.

Molecular disease in plasma vs leukocytes. (A) By analyzing cellular and plasma fractions isolated from 82 samples acquired from 25 patients, we found that clonotype was more abundant per million genome equivalent (G.E.) in the plasma fraction. This included 2 types of cellular fractions: peripheral blood mononuclear cells (blue circles) and plasma-depleted whole blood (purple squares). For 11 samples from 7 different patients, tumor clonotype was detected solely within the plasma fraction and not the cellular fraction, including 2 samples taken at the time of CNS relapse. A total of 7 samples from 4 patients had the tumor clonotype detected only in the cellular fraction and not in the plasma. (B) Thirty-four paired samples with both plasma Ig-HTS and LDH and (C) 26 paired samples with both cellular Ig-HTS and LDH at time points of overt disease (diagnosis or progression/relapse). Plasma Ig-HTS was significantly more sensitive than LDH (88% vs 59%, P = .01). Cellular Ig-HTS was not more sensitive than LDH (62% vs 59%, P = 1.0). (D) Detection of molecular disease in the plasma fraction of blood was significantly correlated with MTV as measured by PET/CT scans (n = 32) at paired time points across 27 patients (P = .002, Pearson r = 0.53). No significant correlation was seen between MTV and cellular Ig-HTS. Plasma samples were taken within 28 days of PET/CT scan (median, 1.5 days).

Ig-HTS monitoring of therapeutic response. Time courses of plasma molecular disease during therapy; red, positive molecular disease; black, negative molecular disease. (A) Patient 26 achieved a CR after primary therapy with DA-EPOCH-R for stage IVE DLBCL, but had progression based on a day 60 PET/CT. Plasma molecular disease at this time point was positive. In the ensuing disease progression during several lines of therapy, disease burden can be seen to increase by both plasma molecular disease and PET/CT. (B) Patient 27 was treated for relapsed stage IVAE DLBCL with DA-EPOCH-R with stable disease shown by PET/CT on day 43. Plasma molecular disease decreased, but remained detectable on days 15 and 43. PD was detected by CT on day 77 (data not shown), corresponding with an increased disease burden by plasma molecular disease that again subsides, but remains detectable, following treatment with R-GDP. (C) Patient 62 was diagnosed with stage IVBE DLBCL including leukemic disease, and disease burden can be seen to decline as the patient achieves a CR. An isolated CNS relapse detected by magnetic resonance imaging on day 250 is also detected by Ig-HTS (day 323).

Limits of disease detection during therapy. (A) Tumor clonotype was detected at varying levels within 1 mL of plasma from 7 of 8 patients before treatment. At an interim time point after 4 cycles of R-CHOP, disease was not detectable in any patient (red circles, detectable molecular disease; gray circles, undetectable molecular disease; red patients, detectable disease on PET/CT; black patients, no detectable disease on PET/CT). (B) PET/CT scans show the disease burden for 8 patients before treatment (left) and after 4 cycles of R-CHOP (right). NHL-75, with no tumor clonotype detected within the plasma at diagnosis, can be seen to have a large tumor burden. Although no patients had detectable disease at the interim time point, 1 case showed residual disease by PET scan. This patient relapsed 376 days later (NHL-11). Two additional cases showed no disease by PET/CT at the interim time point but went on to relapse 247 (NHL-75) and 267 days (NHL-68) later.

Ig-HTS surveillance for disease progression/relapse (A) Patient 41 was treated for stage 1A DLBCL with R-CEOP, achieved a CR, and was monitored thereafter. Plasma ctDNA was positive on day 441 within the plasma, before detection by PET/CT and diagnosis of PD (lead time = 88 days). (B) Patient 46, diagnosed with stage IIA DLBCL, was monitored during treatment encompassing 2 separate relapses. We observed a rapid decline in plasma disease to undetectable levels as this patient achieves a partial response. On days 42 and 84, residual disease was detected by PET/CT (data not shown) despite negative plasma Ig-HTS. A subsequent PET/CT detected progression (day 140) that was also detected by Ig-HTS. The patient achieved a CR by PET/CT after salvage chemotherapy on day 223, but ultimately relapsed as detected by plasma molecular disease on day 473, confirmed by PET/CT on day 487 (lead time = 14 days). (C) Patient 48 was diagnosed with stage IVBE DLBCL with leukemic phase and bone marrow involvement. After treatment with R-CHOP, the patient achieved a CR by flow cytometry on day 228 (data not shown), at which time disease was undetectable Ig-HTS. Plasma molecular disease was detected at a level of 71 860 molecules per million genome equivalents on day 493, despite no evidence of disease on peripheral blood flow cytometry on day 493 or PET/CT (data not shown). Plasma Ig-HTS was positive 162 days before the detection of PD by flow cytometry on day 655. (D) Patient 52 was initially expectantly observed with stage IIIA indolent follicular lymphoma, at which point plasma molecular disease was negative despite clear radiographic evidence. This patient then underwent clinical and histological transformation to DLBCL in a single site that coincided with a dramatic spike in plasma molecular disease. He was treated with R-CHOP chemotherapy for DLBCL, with subsequent CR on PET/CT and normalization of plasma molecular disease.

Superior specificity of Ig-HTS compared with PET/CT. (A) A comparison of the sensitivity and specificity of Ig-HTS and PET/CT in patients who had both tests performed during clinical surveillance, with the performance assessed by ultimate clinical outcome of the patient. PET/CT can be seen to have superior sensitivity to Ig-HTS, but the specificity of Ig-HTS was 100% and greater than that of PET/CT. (B-D) Examples of false-positive PET/CT scans with concurrent true-negative Ig-HTS. (B) A PET/CT of a patient in remission from stage IV DLBCL showed a positive signal at the site of previous disease (left, white arrow). Biopsy showed unremarkable fibro-adipose tissue. (C) A patient in remission from stage IV DLBCL with a positive PET/CT in the lungs (left) resulting from granulomatous inflammation (right). (D) A PET/CT of a patient in remission from stage IV DLBCL with signal in a para-aortic lymph node (left) that was shown to result from granulomatous lymphadenitis (right).