Level of Clonal Plasma Cells in Hematopoietic Cell Autografts Reflects the Pre-Transplant Bone Marrow Minimal Residual Disease in Multiple Myeloma Patients

Abstract

High-dose melphalan followed by autologous stem cell transplantation (ASCT) remains the standard of care for multiple myeloma (MM) patients. However, hematopoietic cell autografts are often contaminated with aberrant plasma cells (aPC) following CD34+ cell mobilization [1], making graft MRD (gMRD) assessment potentially valuable. Nevertheless, the clinical significance of gMRD was not fully leveraged so far. It was demonstrated that the gMRD positivity is associated with worse progression-free survival (PFS) and overall survival (OS) [1, 2]. Recent studies utilizing next generation flow cytometry (NGF) with the limit of detection (LOD) 0.0002% (2 × 10⁻⁶) provided similar results, together with predicting worse post-induction response in gMRD positive patients [3-5]. Therefore, gMRD evaluation holds significant potential for patient benefit. However, there is currently virtually no knowledge about the relationship between gMRD status and pre-ASCT MRD levels in the bone marrow (BM), despite the invasive MRD assessment is increasingly common at this timepoint due to improved first-line therapeutic regimens that induce rapid and deep responses [6].

To enhance the understanding of gMRD clinical significance, we performed a single-center study involving 99 transplant-eligible MM patients diagnosed between 2019 and 2023. All patients were treated with standard induction therapy, followed by CD34+ cell mobilization, leukapheresis of hematopoietic cell graft, high-dose melphalan, and ASCT (Methods S1–S3). MRD was evaluated in grafts on a day of the first apheresis (gMRD), in BMs one day prior to ASCT (pre-ASCT), and in BMs day +100 after ASCT (post-ASCT) using NGF by EuroFlow protocol (Methods S4, Figure S1).

In total, 44% (44/99) of patients in our cohort were gMRD+ and 56% (55/99) were gMRD-. Both groups were well balanced without any significant differences in terms of age, gender, cytogenetic risk, induction regimens used or number of their cycles (median number of cycles = 4), mobilization regimen, or maintenance status. Notably, patients in the gMRD- group more frequently received anti-CD38 therapy (22% vs. 6.8%), whereas VTd was more common in the gMRD+ group (64% vs. 49%); however, the difference was not statistically significant. ISS I stage was more prevalent in gMRD- group (ISS I, II, III frequency: 47%, 31%, 22%), while ISS II and III were more frequent in gMRD+ cohort (18%, 43%, 39%; p = 0.009). R-ISS comparison showed similar results (R ISS I, II, III frequency: gMRD- 44%, 51%, 5.5%, gMRD+ 19%, 69%, 12%; p = 0.033). CD34+ cell yields were significantly higher in gMRD+ group (median, range CD34+ cells × 10⁶/kg: 13, 9–24 vs. 10, 6–14, p = 0.013), which might be partially attributed to higher rates of AraC + G-CSF administered in this group (30% vs. 13%), although the difference was not significant (p = 0.11). Tandem ASCT was more frequent in gMRD+ group (41% vs. 22%; p = 0.04). Plerixafor was administered in 12% (12/99) of patients, without any significant difference between gMRD+ and gMRD- groups. Melphalan dose was reduced in 17% (17/99) of patients without any significant effect on post-transplant MRD status.

As expected, significant differences were observed between the depth of post-induction response and gMRD status. Complete remission (CR) and very good partial response (VGPR) were achieved more frequently in the gMRD- group compared to the gMRD+ group, while partial response (PR) was more common in the gMRD+ group (CR: 31% vs. 9%, VGPR: 51% vs. 34%, PR: 15% vs. 52%; p = 0.001) (Figure S2). Patient characteristics and clinical data are summarized in Table S1. Next, we aimed to confirm the prognostic value of gMRD status assessed by NGF. Although the median PFS for the entire cohort was not reached, the 2-year PFS was significantly prolonged in gMRD- patients compared to gMRD+ patients (82% vs. 64%; p = 0.043) (Figure 1A) with median follow-up since gMRD assessment being 29.9 months. Testing the 2-year PFS between the gMRD+ and gMRD- groups within selected pre-transplant response categories (CR, VGPR or better, VGPR, and PR) to evaluate the additional prognostic value of gMRD did not yield significant results (Figure S3). The prognostic value of pre-ASCT serological response was not significant.

FIGURE 1

Open in figure viewerPowerPoint

Predictive Value of MRD Assessment in Autografts of Multiple Myeloma Patients. (A) The Kaplan–Meier 2-year PFS curves demonstrate worse PFS in gMRD+ MM patients compared to the gMRD- group (p = 0.043). (B) Illustration of proportions of MRD results in autografts and pre-ASCT BM (C) Spearman correlation between gMRD levels and pre-ASCT BM MRD levels revealing a strong association in aPC infiltration levels (r = 0.74, p < 0.001). (D) Paired box plots show MRD levels in autografts and pre-ASCT BM for both gMRD+ (red dots and lines) and gMRD- (blue dots and lines) groups. A significant increase, with a median 2-log difference in aPC infiltration (p < 0.001), can be seen between autografts and pre-ASCT BM in the gMRD+ cohort (red). Additionally, a 2-log difference (p < 0.001) is noted also in pre-ASCT BM between the gMRD+ (red box on the right) and gMRD- groups (blue box on the right). NGF LOD 0.0002% is represented by the dotted line.

Subsequently, MRD levels assessed by NGF were compared in grafts (n = 99), paired pre-ASCT BMs (n = 76), and paired post-ASCT BMs (n = 92). MRD positivity was observed in 79% (60/76) of patients in the BM pre-ASCT (median MRD level: 0.1385%) and in 61% (56/92) post-ASCT (median: 0.0209%). The NGF threshold LOD of 0.0002% was achieved in all gMRD samples due to consistently high cell counts. The median LOD 0.0002% was reached in both pre-ASCT (range: 0.0002%–0.0033%) and post-ASCT BM (0.0002%–0.0015%) samples. In the gMRD+ group, the median level of aPC infiltration was 0.0076% (0.0003%–0.79%). MRD levels of ≥ 10⁻⁴, < 10⁻⁴/≥ 10⁻⁵, and < 10⁻⁵/≥ 2 × 10⁻⁶ were detected in 48%, 32%, and 20% of positive grafts, respectively.

The comparison of gMRD status with paired BM MRD assessed pre-ASCT (n = 76) and post-ASCT (n = 92) revealed clinically highly relevant results. All patients tested gMRD+ were also MRD+ in the pre-ASCT BM (100%, 32/32), while gMRD- patients were in pre-ASCT BM either MRD+ in 64% (28/44) or MRD- in 36% (16/44) (Figure 1B, Tables S2 and S3). Moreover, 83% (34/41) of gMRD+ patients remained MRD+ after ASCT (Table S3). Subsequently, a strong correlation was observed between gMRD levels of all paired grafts and pre-ASCT BMs (r = 0.74, p < 0.001) (Figure 1C), demonstrating a median 2-log increase in MRD level (of MRD+ samples) detected in pre-ASCT BM compared to paired grafts (1.1%, 0.0128%–11.7% vs. 0.01%, 0.0003%–0.79%; p < 0.001). Additionally, the gMRD+ group exhibited a median pre-ASCT MRD level that was also 2 logs higher than that of the gMRD- group (1.1% vs. 0.021%; p < 0.001) (Figure 1D).

Furthermore, to assess the potential impact of gMRD+ rates across different LODs, we performed analyses with the LOD for gMRD assessment artificially set at 10⁻⁵ and 10⁻⁴. Rates of gMRD+ were LOD 10⁻⁵ = 33% and LOD 10⁻⁴ = 19% (vs. LOD 2 × 10⁻⁶ 44%). To further illustrate the impact of false-negative gMRD results on PFS, the Kaplan–Meier curves were also analyzed for LODs of 10⁻⁵ and 10⁻⁴ showing more significant results compared to LOD 2 × 10⁻⁶ in our dataset (Figure S4).

To the best of our knowledge, this is the first study focused on the comparison of gMRD and pre-ASCT BM MRD assessments using a standardized NGF protocol. Our analysis confirmed the prognostic value of gMRD positivity and its association with shorter PFS. Moreover, we demonstrated a high value of gMRD positivity for predicting pre-ASCT BM MRD positivity. In fact, all patients with detectable gMRD were invariably positive for MRD in their pre-ASCT BM samples with a relatively consistent 2-log difference between MRD levels in grafts and pre-ASCT BMs. This finding suggests that gMRD assessment may provide a reliable estimate of both pre-ASCT BM MRD status and level, offering a minimally invasive alternative to traditional BM aspiration. Indeed, according to our results the invasive pre-ASCT evaluation of BM MRD might be omitted in gMRD+ patients. Presented data, combined with results from other studies assessed using NGF, suggest that at least 40% of patients deemed gMRD+ could avoid BM aspiration prior to ASCT. In detail, our cohort discovered 44% (44/99) gMRD+ rate (median LOD 0.0002%; 2 × 10⁻⁶), while when simulating LODs of 0.001% (10⁻⁵) and 0.01% (10⁻⁴), the gMRD+ rates decreased to 33% and 19%, respectively. Kostopoulos et al. reported 40% of gMRD+ patients (79/199; LOD 0.00036% [3.6 × 10⁻⁶]) [4]. Urushihara et al. reported a gMRD+ rate of 71% (35/49) with a LOD as high as 0.00005% (5 × 10⁻⁷) [5]. Study by Pasvolsky et al. represents the largest dataset to date, nonetheless, its results indicate a high rate of false-negative MRD in grafts due to LOD of only 0.05%–0.001% (5 × 10⁻⁴ to 1 × 10⁻⁵), with 18% (75/416) gMRD+ patients [3].

It is reasonable to expect that patients with higher ISS and R-ISS scores may have poorer responses, which could be associated with higher gMRD levels. The clinical implications of our findings are particularly relevant in the context of modern therapeutic strategies utilizing regimens containing CD38 targeting antibodies where deep responses are increasingly common, potentially requiring post-induction MRD evaluation in higher number of patients [6]. Also, highly effective induction regimens will inevitably lead to lower levels of gMRD, making graft assessment with LOD at least on the level of NGF crucial to minimize the rate of false-negative results. Moreover, increasing NGF sensitivity in gMRD, as demonstrated by Urushihara et al. [5] who utilized a modified NGF with LOD 0.00005% (5 × 10⁻⁷), seems to be feasible due to high cell counts in hematopoietic cell grafts. From a practical perspective, gMRD represents an easily accessible, reliable, and minimally invasive surrogate for pre-ASCT BM MRD assessment, potentially eliminating the need for BM aspiration in a large subset of transplant-eligible patients and significantly reducing the procedural burden.