Immunomodulation in Pomalidomide, Dexamethasone, and Daratumumab-Treated Patients with Relapsed/Refractory Multiple Myeloma

Clinical outcomes for patients with multiple myeloma improved dramatically with the advent of novel therapeutic approaches, including, immunomodulatory agents, proteasome inhibitors, and mAbs (1, 2). While pomalidomide (POM) and low-dose dexamethasone (LoDEX) is a safe and efficacious combination for patients with relapsed/refractory multiple myeloma (RRMM), triplet regimens are being explored to improve patient outcomes and fulfill segments of unmet need. Clinical benefit of pomalidomide, LoDEX, and daratumumab (DARA), even in patients with prior exposure to pomalidomide- or daratumumab-based regimens (3–5), highlight the need to understand the mechanisms of enhanced activity between these agents.

Thalidomide analogues exert dual tumoricidal and immunomodulatory activities. Pomalidomide and lenalidomide bind cereblon, part of an E3 ubiquitin ligase complex, and induce polyubiquitination and degradation of substrate proteins, Ikaros and aiolos (6, 7), and transcription factors regulating immune cell development and homeostasis (8, 9). In vitro studies with pomalidomide or lenalidomide resulted in broad costimulatory effects in primary human T and natural killer (NK) cells including proliferation, cytokine secretion, and cytotoxicity (7, 9, 10). In vivo, pomalidomide treatment led to immunomodulation of innate and adaptive immunity correlating with clinical antitumor effects (11).

Daratumumab is a fully human antibody against CD38, expressed on the surface of multiple subsets of immune cells, including myeloma plasma cells (12). Daratumumab is approved in combination with dexamethasone and with pomalidomide and dexamethasone for patients with relapsed/refractory myeloma who have received lenalidomide and a proteasome inhibitor (13). Daratumumab is proposed to exert immunomodulatory effects through depletion of CD38-expressing regulatory B and T cells (14) and NK cells (15).

The direct antitumor activity of pomalidomide along with immunomodulatory activity overcomes immunosuppression by dexamethasone (11). Given the complex nature of daratumumab immune activity, we sought to understand pharmacodynamics of this triplet regimen in patients with RRMM. To study the rationale for combination, longitudinal immune profiling of patients with multiple myeloma on pomalidomide, LoDEX, and daratumumab was conducted. Results were consistent with the previously identified negative impact on total NK cells by daratumumab (16, 17), but here, counterbalanced with enhanced NK proliferation (Ki67⁺) due to POM-DARA-LoDEX combination. Furthermore, this combination may increase T-effector memory (EM) cell proliferation/activation without increase in T-cell exhaustion or regulatory T cell (Treg) subpopulations.

MM-014 is a phase II, multicenter, two-arm nonrandomized [POM-LoDEX (arm A; ref. 18) or POM-loDEX-DARA (arm B) treatments], open-label clinical trial conducted at multiple study sites (ClinicalTrials.gov NCT01946477). The analyses conducted in this study were focused on the arm B cohort. Study objectives and patient treatment are described in Supplementary Data. Baseline patient characteristics of the immune biomarker subgroup (n = 110) were similar to those of the intent-to-treat population (n = 112; Supplementary Table S1). Notably, median number of prior antimyeloma regimens was two and all patients received prior lenalidomide treatment.

This study was approved by each site's institutional review board or ethics committee. All patients provided written informed consent. The study was executed in accordance with the principles of the Declaration of Helsinki. The investigators designed the study in conjunction with the sponsor, Bristol Myers Squibb.

Whole-blood samples were collected at indicated timepoints in Na-Heparin, tubes per protocol (Supplementary Fig. S1). All analyses were conducted by Q2 Solutions with exception of flow cytometric evaluation of CD38 expression, which were conducted at BARC USA. For analyses conducted at Q2 Solutions, lymphocyte and monocyte frequencies were examined by flow cytometry using validated assays. Initial gating included removal of the debris (FSC-A^low events) and gating on singlet FSC and SSC populations followed by lymphocytes gating on CD45⁺SSC low cells. Lymphocyte gate was adjusted to include T cells, B cells, and NK cells and to exclude CD45^dimCD3⁻CD19⁻CD16/56⁻ cells. Antibodies and conjugated fluorophores used are provided in Supplementary Materials and Methods.

Descriptive statistics were generated using Prism (GraphPad software) and SAS. Change from baseline for each pharmacodynamics parameter was calculated as a difference between the last available postbaseline data and baseline data. Baseline was calculated as an average of numeric result in standard units in screen and cycle (C) 1 day (D) 1 timepoints. Statistical analysis for significant association with response was preformed using Wilcoxon rank-sum test. Following designations were used to indicate significance levels: *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

We examined immune composition of major cell types (NK, CD4⁺ T, CD8⁺ T, Treg, B cells, and monocytes) by absolute counts and percentage lymphocyte gate at baseline and during treatment. Baseline immune phenotyping showed a mean of 36.2% CD4⁺ T cells, 28.6% CD8⁺ T cells, 20.2% NK cells, and 11.9% B cells in patients. Potent effects of treatment on immune composition were observed in samples from treated patients. Most notably, NK cells were depleted by approximately 80% (P < 0.001) following the first daratumumab administration and remained depleted throughout observation (Fig. 1A; Supplementary Tables S2 and S3). Peripheral CD19⁺ B cells were also significantly depleted (P < 0.001), and B-cell numbers and frequencies remained below baseline during treatment. In contrast, peripheral T cells (by counts and relative frequencies) increased starting on C1D8 and remained consistently above baseline levels during treatment (Fig. 1A; Supplementary Tables S2 and S3). Treatment-induced depletion of immune subsets appeared to correlate with CD38⁺ expression. CD38⁺ expression was observed in a mean of 72.7% of NK cells and 31.3% of B cells versus only 11.2% of CD8⁺ and 9.2% of CD4⁺ T cells expressing CD38⁺ (Fig. 1B). Sustained depletion (>80%) following treatment was, however, noted in CD38⁺CD3⁺CD4⁺ and CD38⁺CD3⁺CD8⁺ T cells (Supplementary Tables S2 and S3). CD38⁺ expression was also observed in 35.9% of CD4⁺CD25⁺CD127dim Tregs. However, treatment did not affect total Treg numbers, likely because they only represented 3.4% of all lymphocytes.

The presence of myeloid-derived suppressor cells (MDSC) has been described in both peripheral blood and bone marrow compartments of patients with multiple myeloma and immunomodulatory compounds have previously been shown to modulate their function (19, 20). We observed a slight trend toward decreased Lin⁻CD14⁺HLA⁻DR⁻/CD33⁺CD11b⁺ M-MDSCs in the peripheral blood of patients treated with the triplet regimen. Although there was an initial surge in Lin⁻CD15⁺/HLA⁻DR⁻/CD33⁺/CD11b⁺ PMN-MDSCs in treated patients, by mid-cycle 1, levels had returned to baseline and remained stable through mid-cycle 2 (Supplementary Fig. S2).

We observed a difference in the magnitude of treatment effect on total versus proliferating NK cells. While absolute NK-cell numbers decreased from baseline by 85.5% at C1D8 in patients (P < 0.001) and remained below baseline throughout treatment, absolute proliferating NK cells only decreased by 24.8% (P = 0.070) and were maintained throughout the study (Fig. 2A). Longitudinal assessment showed that the proportion of NK cells proliferating (%Ki67⁺) increased significantly from approximately 10% at baseline to 46.2% (P < 0.001) by C1D8 (Fig. 2B). While proportion of Ki67⁺ NK cells remained above baseline throughout treatment, they significantly decreased between C1D15 and C2D1 (P < 0.001), after a 7-day pomalidomide holiday, and then, conversely, increased by 35.2% (P < 0.001) between C2D1 and C2D15, after pomalidomide dosing was resumed. Although the contribution of the individual agents is difficult to dissect in the absence of daratumumab-only or pomalidomide-only treatment arms, these results are consistent with pomalidomide-mediated proliferation of NK cells in presence of daratumumab.

Pomalidomide has previously been described ex vivo as a costimulator of T-cell activation and proliferation (7, 9). Consistent with these studies, we observed a significant increase in both the frequency and absolute number of proliferating CD3⁺ T cells in patients on C1D8 by more than 2-fold (P < 0.001). Mirroring our observations on induction of NK-cell proliferation, the proportion of proliferating T cells significantly (P < 0.001) decreased at C2D1, after a 7-day pomalidomide holiday, and then rebounded at C2D15, after pomalidomide dosing was resumed, increasing by 67.1% (P < 0.001; Fig. 3A; Supplementary Tables S2 and S3). Notably, this proliferative response was more pronounced in CD8⁺ than in CD4⁺ T cells (Fig. 3B). Analogous to changes in T-cell proliferation, the proportion of activated (HLA-DR⁺) CD4⁺ and CD8⁺ T cells also increased from baseline to C1D8 by 44.7% (P < 0.001) and 22.0% (P < 0.001), respectively (Fig. 3C). Longitudinal analysis revealed that HLA-DR positivity was also significantly reduced in both CD4⁺ and CD8⁺ T cells on C2D1, after 7-day pomalidomide holiday, and then increased again, after pomalidomide dosing was resumed, demonstrating a cyclical pattern of immunomodulation associated with pomalidomide dosing (Fig. 3C).

We examined whether treatment-driven enrichment of proliferating and activated T cells led to a shift in T-cell pool composition by analyzing naïve (CD45RA⁺CD45RO⁻CCR7⁺), central memory (CM, CD45RA⁻CD45RO⁺CCR7⁺), and EM (CD45RA⁻CD45RO⁺CCR7⁻) CD4⁺ and CD8⁺ T-cell subsets (21). We observed that proportion of naïve CD4⁺ and CD8⁺ T cells (percentage of total CD4⁺ and CD8⁺ T cells, respectively) was reduced by 50%–75% at C2D15 compared with the screening timepoint (Fig. 4A; Supplementary Tables S2 and S3). In addition, central memory T cells (Tcm) cells remained near or below baseline levels (Fig. 4A). In contrast, a strong trend toward increased EM T cells was observed with significantly elevated EM CD8⁺ T-cell numbers on C2D15 relative to screening (Fig. 4A; Supplementary Tables S2 and S3). Notably, the proportion of both CD4⁺ and CD8⁺ EM T cells was significantly increased at all on-treatment timepoints examined (Fig. 4B). Consistent with previous observations indicating a cyclical pattern of immunomodulation associated with pomalidomide dosing, trends in proportion of EM cells decreased toward baseline after 7-day pomalidomide holiday on C2D1, and hen increased again on C2D15 (Fig. 4B).

Analysis of Tregs indicated that CD38-expressing Tregs were depleted in treated patients, presumably due to anti-CD38⁺ cell activity of daratumumab (14). Previously noted pomalidomide-induced expansion of activated and EM T cells (22) may be further augmented by depletion of this population, contributing to enhanced benefit with daratumumab. Although approximately 40% of CD25⁺ CD4⁺ Tregs expressed CD38 at baseline (Fig. 4C), we previously noted that Tregs frequency mostly remained unchanged. In addition, minimal changes in frequency of CD45RO⁺ (previously activated and/or proliferating Tregs) or CD45RA⁺ (naïve) Treg cells (Fig. 4C) were observed. Taken together, we observed T-cell differentiation and proliferation, occurring without concomitant increase in Tregs, which would be expected to positively impact clinical benefit of the triplet regimen.

Extensive and continuous T-cell stimulation and differentiation may result in increase of cells with exhausted phenotype and expression of coinhibitory receptors and plays a role in cancer progression (23, 24). We investigated whether the sustained T-cell proliferation and activation following administration of POM-LoDEX-DARA drove T cells toward exhaustion. We saw no increase in the expression of single receptor [LAG-3 (CD223), PD-1 (CD279), or TIM3 (CD336)]-expressing CD4⁺ or CD8⁺ T cells during treatment (Supplementary Fig. S3). These results suggest that T-cell activation and costimulation did not result in emergence of T cells displaying exhausted phenotype in the first two cycles.

Finally, we analyzed pharmacodynamic immune responses in lenalidomide-refractory and lenalidomide nonrefractory patients (Supplementary Tables S4 and S5) and observed potent immune modulation in both patient subgroups that were largely overlapping. For example, proliferating NK cells increased 4.7- and 4.6-fold relative to baseline and percentage changes increased 362% and 368%, respectively. Consistent trends between lenalidomide-refractory and non-lenalidomide–refractory patients were also noted in increased proliferating, HLA-DR⁺, ICOS⁺, and EM T-cell subsets, and decreased naïve T cells on-treatment (Supplementary Tables S4 and S5). These results show POM-DARA-loDEX treatment can stimulate the immune system in patients refractory to a previous line of immunomodulatory agent. Presumably, this immunostimulation could contribute to observed clinical activity of the triplet combination.

Viably preserved pre- and posttreatment bone marrow aspirate bone marrow mononuclear cells were assessed by mass cytometry in 6 patients (CyTOF; panel markers in Supplementary Table S6) and compared with peripheral immune changes in these same patients with consistent trends in pharmacodynamic changes observed. As noted previously, the most profound impact on immune composition was NK-cell decrease (83% in microenvironment vs. 84% periphery), consistent with anti-CD38⁺ NK-cell activity of daratumumab (Supplementary Table S7). Tumor microenvironment (TME) showed a decrease of 51% in CD38⁺ T cells versus 90% peripheral loss. Consistent trends in TME relative to periphery were noted by increase in bone marrow monocytes (52.0%) and periphery (60.8%). B cells were decreased in periphery (64%) and bone marrow (52%). Consistent trends were also noted in naïve T cells decrease in periphery (CD4⁺, −69% and CD8⁺, −74%) and bone marrow (75% and 70%, respectively). An overall shift in T-cell pool favoring increased CD8/CD4 ratio in memory cell population was observed in both compartments. Single-cell unsupervised analyses confirm these trends (Supplementary Fig. S4).

Immune cell phenotyping at baseline and change from baseline at C2D15 were assessed for association with progression-free survival (PFS, nonresponder/relapsed vs. nonrelapsed patients) and response [≥very good partial response (VGPR) vs. <VGPR]. At 12 months PFS, there were few relapse events and at 24 months, many patients remain censored, hence we focused our analyses on 18-month landmark PFS. At 18 months PFS, there was significant association of median baseline absolute cell counts between nonrelapsed and relapsed patients for CD3⁺CD4⁺ (420 cells/μL vs. 278 cells/μL, respectively; P = 0.030). Higher baseline and posttreatment CD4⁺ T-cell counts were also associated with response ≥VGPR (P = 0.002 and P = 0.015, respectively). Other CD4⁺ subsets were also associated with 18-month PFS, specifically, CD4⁺ naïve T-cell counts (120 cells/μL vs. 64 cells/μL; P = 0.045), CD4⁺ Tcm (188 cells/μL vs. 117 cell/μL; P = 0.031), and activated CD4⁺ICOS⁺ (295 cells/μL vs. 196 cells/μL; P = 0.015). A trend toward PFS association was observed with CD4⁺CD38⁺ Tregs (267 cells/μL vs. 0.163 cells/μL; P = 0.056). Consistent with the role of baseline and posttreatment CD4⁺ subset counts with outcomes, the following were also significant for ≥VGPR patients: CD4⁺ Tcm (P < 0.001 and P = 0.001), CD4⁺ICOS⁺ (P = 0.014 and P = 0.017), and CD4⁺CD38⁺ (P = 0.021 and P = 0.025). No significant associations with 18-month PFS were observed for CD8⁺ T cells or subpopulations [although baseline and posttreatment CD8⁺ Tcm was associated with response (P = 0.0061 and P = 0.022, respectively)]. CD4⁺ naïve T cells at baseline were not associated with ≥VGPR, but posttreatment measures were significant (P = 0.014). Trends were noted in baseline levels of CD3⁻CD16⁺CD56⁺ NK cells with 18-month PFS (250 cells/μL vs. 140 cells/μL; P = 0.013). Furthermore, no PFS outcome associations were noted in CD19⁺ B cells, CD14⁺ monocytes, CD4⁺FoxP3⁺ Tregs, or exhausted T-cell counts at baseline or changes posttreatment at C2D15 (all P > 0.05).

Immune subsets at baseline found to be significantly associated with 18 months landmark PFS above were further analyzed by significance in PFS difference between immune counts, stratified by median (high vs. low). Kaplan–Meier plots in Fig. 5 show that median-high CD4⁺ cell counts displayed a median PFS of 21.9 months [95% confidence interval (CI), 18.7–NA] compared with median-low counts at 15.9 months [95% CI, 9.7–NA; HR, 0.46 (95% CI, 0.23–0.92), P = 0.025]. In addition, CD4⁺ICOS⁺ median-high counts had a PFS of 21.9 months (95% CI, 18.7–NA) versus 15.9 months (95% CI, 9.7–NA) for median-low [HR, 0.50 (95% CI, 0.25–1.02), P = 0.051]. Significance for CD4⁺ and strong trend for CD4⁺ICOS⁺ with PFS was maintained following covariate analyses with clinical adjustment variables (Supplementary Table S8). Of the 100 patients comprising this set of analyses, 67 remain censored and so the statistical analysis will continue to be followed, especially for correlations of posttreatment changes in addition to baseline measures.

We performed extensive immunophenotypic analyses on longitudinal peripheral blood samples from a uniformly treated large cohort of patients enrolled in a clinical study of POM-loDEX-DARA. Our analyses confirm and extend previous observations on the putative mechanisms of both pomalidomide and daratumumab. Pomalidomide mediated increases in proliferating T cells, ICOS⁺, and HLA-DR⁺ activated T cells, and expansion in EM T-cell compartment with strongest positive impact on cytolytic CD8⁺ T cells. Toward compensatory activity, loss of NK cells and CD38-expressing T cells owing to daratumumab was offset by an increase in pomalidomide-driven NK-cell proliferation and increases in proliferating and activated T cells. While immunomodulatory agent treatment has been shown to increase Tregs, here, combination treatment also resulted in stable CD4⁺/FoxP3⁺ Tregs, which would be expected to positively impact clinical benefit of the triplet regimen. Finally, we also analyzed immune profiling posttreatment for lenalidomide-refractory patients and showed that immune modulation mirrors that of total patients population. Importantly, these results indicate that POM-loDEX-DARA could promote increases in the above indicated immune subpopulations in patients refractory to previous immunomodulatory agent treatment.

In recent studies, POM-DEX treatment of patients with RRMM induced increase in T- and NK-cell populations and expansion of activated and EM T cells (11, 22). In MM-014 arm A patients with RRMM, administered with POM-loDEX (receiving lenalidomide in most immediate prior line of therapy with 87% refractory), an expansion in CD4⁺ and CD8⁺ T cells was also observed (18). In this study, daratumumab was administered once per week and has an 18-day half-life, indicating that it effectively had no drug holiday and exposures are steady state through the end of C1. In contrast, pomalidomide was given once per day for 21 days of the 28-day cycle and thus, experienced a drug holiday of 1 week. Diminution and lack of recovery of CD16⁺CD56⁺ NK cells observed in our study occurred concurrently with daratumumab administration and exposure. In contrast to this observation, CD3⁻CD19⁺ B cells dynamics appeared to be driven by both pomalidomide and daratumumab exposure. B cells were initially depleted, but rebounded during pomalidomide holiday, before diminishing again once pomalidomide treatment was resumed. This suggests a pomalidomide-driven role for the anti-B-cell properties of the triplet therapy.

Similarly, we also observed changes in activated (HLA-DR⁺), proliferating (Ki67⁺), naïve, and EM T cells that were reversed during drug holiday and recovered when pomalidomide treatment resumed. It is challenging to discern changes by individual components of a treatment regimen in which three agents are administered. In vitro treatment of healthy volunteer PBMCs with daratumumab, pomalidomide, or combination and assessment of changes in immunophenotyping showed increases in total and Ki67⁺ proliferative NK cells with pomalidomide-treated stimulated PBMCs. Conversely, treatment with daratumumab alone resulted in total NK cells decrease and little change in Ki67⁺ proliferative NK cells. Notably, the combination showed stable NK-cell frequency and proliferating NK cells increase. Single-agent pomalidomide favored an increase in CD8⁺ T cells, appreciably proliferating CD8⁺ cells in the presence of both pomalidomide alone and in combination with daratumumab, but not in daratumumab alone (Amatangelo, unpublished observations). While it has been previously shown that pomalidomide may enhance cytokine production (9, 11, 25), and here we have described enhanced changes in HLA-DR⁺ and ICOS⁺ T cells, along with Ki67⁺ proliferation, additional attention should be directed toward activation state and functional changes to these now established therapeutic properties.

Finally, we found an association of baseline peripheral immune characteristics associated with 18-month landmark PFS. Interestingly, significant association appeared to draw from CD4⁺ T cells, where higher levels of total CD4⁺, naïve CD4⁺, and ICOS⁺ CD4⁺ T cells were associated with patients not progressed at 18 months. Association of higher total NK cells at baseline was also associated with improved outcome. When high versus low immune cells were examined by median, favorable trends were noted in improved PFS in high CD4⁺ and CD4⁺ICOS⁺ immunophenotypes. No statistical associations between pharmacodynamic changes in peripheral immune characteristics were observed, although these data will continue to mature; currently 67% of patients remain censored for PFS.

In summary, we examined increases in immune cell subpopulations that may emerge from combination of separate therapeutic modalities in the treatment of relapsed myeloma. These increases in immune subpopulations were exhibited in lenalidomide-refractory patients highlighting that POM-DARA-loDEX may positively impact immunity in settings where earlier lines of immunomodulatory agents have failed. By employing novel cereblon-modulating agents along with additional emerging antibody-based therapies, such combinations may drive enhanced immune response along with cell autonomous direct tumoricidal activity.

W.E. Pierceall reports other from Bristol Myers Squibb (employment/equity ownership) during the conduct of the study. M.D. Amatangelo reports other from Bristol Myers Squibb (employment and equity ownership) during the conduct of the study and outside the submitted work. N.J. Bahlis reports other from Celgene/Bristol Myers Squibb (advisory board member), grants and other from Janssen (advisory board member), other from Takeda (advisory board member), Karyopharm (advisory board member), Amgen (advisory board member), Sanofi (advisory board member), GlaxoSmithKline (advisory board member), Genentech (advisory board member), and Pfizer (advisory board member) outside the submitted work. D.S. Siegel reports other from Celgene (speakers bureau, advisory boards), Bristol Myers Squibb (speakers bureau, advisory boards), Amgen (speakers bureau, advisory boards), and Janssen (speakers bureau, advisory boards) outside the submitted work. A. Rahman reports grants from Bristol Myers Squibb during the conduct of the study. M. Young reports other from Bristol Myers Squibb (employment/equity ownership) during the conduct of the study. S. Parekh reports grants from Celgene during the conduct of the study, personal fees from Foundation Medicine, and grants from Karyopharm outside the submitted work. A. Agarwal reports other from Bristol Myers Squibb (employment) during the conduct of the study, other from Bristol Myers Squibb (employment) outside the submitted work. A. Thakurta reports other from Bristol Meyers Squibb (employment/equity ownership) during the conduct of the study. No potential conflicts of interest were disclosed by the other authors.

W.E. Pierceall: Conceptualization, data curation, formal analysis, supervision, visualization, writing-original draft, writing-review and editing. M.D. Amatangelo: Formal analysis, visualization, writing-original draft, writing-review and editing. N.J. Bahlis: Resources, data curation, formal analysis, methodology, writing-review and editing. D.S. Siegel: Resources, formal analysis, investigation, writing-original draft, writing-review and editing. A. Rahman: Resources, formal analysis, methodology, writing-original draft, writing-review and editing. O. Van Oekelen: Formal analysis, visualization, writing-original draft, writing-review and editing. P. Neri: Resources, supervision, investigation, writing-original draft, writing-review and editing. M. Young: Formal analysis, visualization, methodology, writing-original draft, writing-review and editing. W. Chung: Formal analysis, visualization, methodology, writing-original draft, writing-review and editing. N. Serbina: Investigation, methodology, writing-original draft, writing-review and editing. S. Parekh: Conceptualization, resources, supervision, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. A. Agarwal: Conceptualization, resources, investigation, writing-original draft, writing-review and editing. A. Thakurta: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, writing-original draft, project administration, writing-review and editing.

This study was funded and sponsored by Bristol Myers Squibb. The study's sponsor compiled and maintained the data. All authors evaluated the results, contributed to the development of the article, and reviewed and approved the article for submission. The study sponsor and all authors accept full responsibility for the accuracy and completeness of the data. The authors acknowledge members of the Thakurta laboratory (appreciably Chad Bjoklund and Patrick Hagner for critical reading of article and helpful comments). Appreciation is extended to Bristol Myers Squibb colleagues Yingdong Lu and Shuai Wang for data interpretation and Faiza Zafar for clinical trial patient information.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.