Minimal residual disease (MRD) detection at end-induction in bone marrow aspirate samples is an independent prognostic tool in childhood and adult B-ALL patients [1-4]. Recently, the concept of MRD detection in peripheral blood samples (PB-MRD) has come to the lime-light [5]. There are a few studies comparing the MRD levels in paired PB and bone marrow (BM) samples collected at varying time points of treatment [5,6]. In addition, some groups have attempted MRD detection in peripheral blood samples at day 15 of induction and have proven its role in early treatment modifications [6,7]. PB-MRD analysis, specifically done on day 15 of induction had better prognostic correlation than the PB-MRD analysis done at later time points (day 33, week 12, pre-maintenance and post-maintenance)[6]. In pediatric B-ALL management, day 15 PB-MRD could be an early and minimally invasive tool to assess treatment response. With this hypothesis, we planned a prospective study to estimate the sensitivity and specificity of mid- induction (day 15) PB-MRD in predicting day 15 BM-MRD.

The patients were enrolled from the department of Pediatrics (Hematology-oncology unit) and the samples were processed and analyzed in the flow cytometry facility of a tertiary-care hospital in India. With institutional ethical clearance, patients were enrolled from July 2012 to May 2013. Under informed consent from the parents, all pediatric B-ALL patients with a diagnostic immunophenotype of bright CD19, CD10 dual expression, and variable (dim to bright) surface CD34 expression (CD19+CD10+CD34+/-) were included. All National Cancer Institute (NCI) standard risk patients underwent three drug induction comprising vincristine, L-asparginase and prednisolone, while the NCI-high risk patients were given daunorubicin in addition. As per the treatment protocol followed in our institute (adapted from UKALL 2003, version 7), a single-routine bone marrow examination was carried out on day 15 of induction (mid-induction) for morphological assessment of treatment response. The first pull bone marrow sample along with a peripheral blood sample were used to evaluate MRD.

The comparison of means between two groups was done using paired t-test or Wilcoxon signed rank test. Comparison of means between more than two groups was done using one way ANOVA and in case of significant result, Bonferroni correction was applied to compare the means of two groups at a time and the significance of each pair of groups was determined. For determination of correlation between two groups Spearman’s rho was used. The tests were considered statistically significant at P <0.05 and were calculated using and SPSS version-17.0 (SPSS Inc., Chicago, Illinois, USA).

Forty patients were enrolled in the study, among which 30 (75%) were NCI standard risk and the rest 10 (25%) were NCI high risk. The cohort had 33 males and 7 females with a male to female ratio of 4.7:1. The mean (SD) age was 5 (2.8) years. All the patients were negative for BCR-ABL1 (p190 and p210), AF4-MLL, ETV6-RUNX1 and E2A-PBX1 by qualitative polymerase chain reaction (PCR).

Of 40 mid-induction BM samples, MRD was positive in 25 samples (62.5%). Of these BM-MRD positive patients, there were 21 males and 4 females with a mean (SD) age of 5.6 (3.0) years. The mean (SD) BM-MRD was 1.17 (1.56) %. The BM-MRD negative patients had a mean (SD) age of 4 (2.1) years comprising 12 males and 3 females. The difference in BM-MRD among the standard risk and high risk patients were insignificant (P=0.20).

Of 40 mid-induction PB samples, 16 (40%) were MRD positive. The PB-MRD positive patients had a mean (SD) age of 5.1 (2.9) years, comprising 14 males and 2 females. The mean (SD) PB-MRD was 0.21 (0.34)%. PB-MRD negative patients comprised 19 males and 5 females with a mean (SD) age of 4.9 (2.8) years. Similar to BM-MRD, the difference in PB-MRD was insignificant (P=0.10) between the standard risk and high risk patients.

There were no statistically significant differences in the baseline and day-15 parameters like, hemoglobin (P=0.39), leukocyte count (P=0.12) and platelet counts (P=0.63), between the BM-MRD positive and negative patients and also between the PB-MRD positive and negative patients (P values of 0.66, 0.49 and 0.97 for hemoglobin, leukocyte count and platelet counts, respectively) (Table I).

TABLE I	Comparison Between Baseline and Day 15 Parameters in Peripheral Blood and Bone Marrow MRD 
Positive and Negative Patients

On analyzing PB and BM as sample pairs, among 40 pairs analysed, 16 sample pairs were MRD positive in both PB and BM. Isolated BM-MRD positivity was seen in 9 sample pairs and 15 sample pairs did not show any MRD in both the samples. However, none of the pairs showed isolated PB-MRD positivity.

Among 25 BM-MRD positive patients, 16 had concurrent PB-MRD positivity (64% of BM-MRD positive patients). The PB and BM-MRD of these 16 patients showed a significant direct correlation (Spearman’s rho=0.670, P=0.007). These 16 PB-MRD positive patients had a mean (SD) BM-MRD of 1.47 (1.84)%, whereas, the mean (SD) BM-MRD in PB-MRD negative patients was only 0.23 (0.52)%. This indicates a 6.3 times higher residual medullary leukemic load in PB-MRD positive patients. Of the PB-MRD positive patients, 75% had BM-MRD of >0.1% and 25% had BM-MRD ranging from 0.01 to 0.1%. A minimum medullary MRD of 0.3% was seen in 81.2% of PB-MRD positive patients. The sensitivity of PB-MRD to identify BM-MRD positivity was 64% with 100% specificity. The positive and negative predictive values were 100% and 62.5%, respectively.

During follow-up, two standard-risk patients died before completing induction and the data of one patient was not traceable. These three patients were day 15 BM-MRD negative. Of the remaining thirty seven patients, the mean (SD) follow-up period was 101(27) weeks. None of the BM-MRD negative patient relapsed. At a mean (SD) of 79 (23) weeks, six BM-MRD positive patients (25%) had relapsed. Three patients had isolated central nervous system (CNS) relapse, one had isolated medullary relapse; concurrent CNS and medullary relapse was seen in one patient while another patient had an initial unilateral testicular relapse at week 51, followed by CNS relapse at week 71. Four out of six patients (67%) with disease relapse had day 15 PB-MRD positivity.

The results of our study reveal that day 15 PB-MRD has a sensitivity of 64% in identifying concurrent day 15 BM-MRD, with 100% specificity. The positive and negative predictive values are 100% and 62.5%, respectively. On follow-up, none of the PB-MRD negative patients had disease relapse, while 67% of relapsed patients had PB-MRD positivity.

This study is limited by the fact that bone marrow examination was not repeated at end of induction phase as the institute’s treatment-protocol used at the time of the study did not have this option, and hence, a comparison of mid-induction and end-induction MRD levels could not be done. In addition, a day 8 blast count by morphology and day 8 MRD analyses were also not attempted. A composite study with day 8, day 15, and day 35 (end-induction) MRD analysis with long term follow up, may yield information on optimal time or combination of time-points for attempting MRD analysis.

The strong relationship between end-induction BM-MRD levels and risk of relapse in childhood B-ALL is well documented [1,3-4,10]. MRD values are used to assess early treatment response, modify treatment intensity and estimate the optimal timing for hematopoietic stem cell transplantation [2,11-13]. In recent past, the concept of PB-MRD has come to lime light. There are few studies comparing the MRD levels in paired PB and BM samples collected at varying time points of treatment [5-6]. In 2002, Coustan-Smith, et al. [5] inferred that in T-ALL, PB-MRD was always positive if BM-MRD was positive (100% concordance) and hence PB-MRD can replace BM-MRD in T- ALL. The concept was not applicable for B-ALL, where the BM and PB MRD levels did not correlate and not all BM-MRD positive patients had concurrent PB-MRD positivity [5]. Similar conclusions were also derived by Van der Velden, et al. [14] in 2002, showing complete concordance between BM and PB MRD in T-ALL, but not in B-ALL [14]. Volejnikova, et al. [6] confirmed that PB-MRD analysis, specifically done on day 15 of induction had better prognostic correlation rather than the PB-MRD analysis done at later time points (day 33, week 12, pre- maintenance and post-maintenance). Authors were of the opinion that defining a low-risk group by PB-MRD negativity as early as day 15 will create an opportunity for de-escalating the therapy [6].

In conclusion, although end-induction BM-MRD presently remains irreplaceable, we suggest that day 15 PB-MRD could be considered as an early, minimally invasive and easily accessible MRD-screening option.

Acknowledgements: Prof Brent L Wood, Director, Hemato-pathology Laboratory, University of Washington, Seattle, USA, for his valuable inputs in standardizing our MRD experiment; and Mrs Kushum Chopra for helping in statistical analysis.

Contributors:MUSS: conceived and designed the study; KBBK: standardized and performed the flow cytometry experiment, collected clinical information, analyzed the data and wrote the manuscript; PB: helped in flow cytometry sample processing; MUS, NV, DB: made critical revisions of the manuscript.

Funding: None; Competing interest: None stated.