1 INTRODUCTION
5.3.1 Mobilization substudy
Table 29. The comparison of all PCR-positive/negative and MFC-positive/negative samples from patients who reached nCR/CR (n=129)
MFC negative, N=91 (%) MFC positive, N=38 (%)
PCR negative 59 (65) 1 (3)
PCR positive 32 (35) 37 (97)
PCR, polymerase chain reaction; MFC, multiparameter flow cytometry; nCR, near complete remission
All of the 129 follow-‐‑up samples from nCR/CR patients were simultaneously investigated by S-‐‑ and U-‐‑ IFE, the S-‐‑FLC ratio, MFC and PCR. Among these samples IFE or S-‐‑FLC ratio did not have clinically predictive correlation with immunophenotypic or molecular remission (Figure 2. II). S-‐‑ or U-‐‑IFE and the S-‐‑FLC ratio were not statistically significantly predictive for MolR (p=0.44 and p=0.06, respectively), and the prediction for MFC-‐‑
negativity was even lower. IFE was still positive in 39% and 38% of MFC-‐‑negative and PCR-‐‑negative samples, respectively. IFE and PCR were concordant in 57% of samples, positive or negative, and the respective number was 62% for IFE and MFC. IFE and FLC ratio were serologically in concordance in 63%. Two patients who were MRD negative by MFC and PCR had consistently positive IFE and an abnormal FLC ratio prior to extramedullary relapse 16 months after ASCT.
The successful comparison of these four methods required an applicable allele specific oligonucleotide for each patient. That would not have been possible without expanding the clonality screening to IgK and IgL multiplex PCRs, IgH singleplex PCRs and reverse oriented ASO primer and probe, which was needed in 13/22 patients and yield 100%
applicability comparable with MFC assay. When MFC was positive, PCR was also positive in 97% of cases.
5.3 RANDOMIZED COMPARISON OF CD34+ CELL MOBILIZATION WITH LOW-‐‑
DOSE CY + G-‐‑CSF OR G-‐‑CSF ALONE AFTER NOVEL INDUCTION TREATMENT (FMG-‐‑MM02) (III)
5.3.1 Mobilization substudy
Low-‐‑dose CY (2g/m2) + G-‐‑CSF 5 µμg/kg (arm A) was compared with G-‐‑CSF 10 µμg/kg alone (arm B), in a randomized fashion in a substudy of the FMG-‐‑MM02 study.
There was no statistically significant difference in the primary endpoint (the proportion of patients with the yield ≥ 3 x 106/kg for one transplant with 1−2 aphereses), 94% vs. 77% in arms A and B, respectively (p=0.084) (Table 30). The only difference was in one of the secondary endpoints. The median number of aphereses needed to reach a yield ≥ 3 x 106/kg was lower in arm A than in arm B [1 vs. 2 aphereses (p=0.035)]. Regarding the predetermined goal of ≥ 6 x 106/kg for a double graft there was no difference between the arms, 62% vs. 51% (p=0.469), respectively. The proportion of patients achieving this goal with 1−2 aphereses was 62% in arm A and 50% in arm B (p=0.662), respectively. Another secondary endpoint, need of plerixafor use, was similar between the two arms; two patients (6%) in arm A and five patients (14%) in arm B (p=0.428).
Even though there was no significant difference between the study arms in terms of B-‐‑
CD34+ cells x 106/l on the first apheresis day there was a significant difference in yields of the first aphereses (Figure 1 and 2, III). On the second day of aphereses both the B-‐‑CD34+
level and the yield were higher in arm A (Figure 1 and 2, III).
Table 30. Mobilization results in the randomized study in MM patients Arm A) CY + G-CSF
N=34 Arm B) G-CSF
N=35 P
Primary endpoint yield ≥ 3 x 106/kg with 1−2 harvests,
N(%) 32/34 (94) 27/35 (77) 0.084
Primary endpoint yield ≥ 6 x 106/kg for double graft with which is the recommended target for a single transplant by IMWG.
The total number of CD34+ cells collected was higher after CY 2g/m2 + G-‐‑CSF than after G-‐‑CSF alone, with median of 6.7 (2.2−12.4) x 106/kg and 5.3 (2.4−12.4) x 106/kg, respectively (p=0.012) (Table 2, III). There was also a statistically significant difference between the arms regarding the number of CD34+ cells infused after high-‐‑dose melphalan; 4.3 (2.2−7.3) x 106/kg and 3.2 (2.3−6.2) x 106/kg in arms A and B, respectively (p=0.010). The engraftment data and hospitalization time were similar between the arms (Table 3, III).
5.3.2 Treatment responses
Sixty-‐‑nine patients were mobilized after three courses of RVD. Altogether 59 patients proceeded to ASCT. The responses before mobilization and three months after ASCT are presented in Table 31 and Figure 14.
The ORR after RVD induction was 84% by intention-‐‑to-‐‑treat (ITT). The primary endpoint, MFC-‐‑MRD negativity measured by 6-‐‑10 color MFC, was achieved after induction in 22/69 (32%) of patients who actually received the assigned RVD-‐‑induction and in 22/80 (28%) by
ITT. The median sensitivity of MFC-‐‑MRD-‐‑negativity was < 0.01% (range 0.0046%−0.069%).
Three months post-‐‑ASCT altogether 24/53 (45%) of evaluable patients were MFC-‐‑MRD negative with a median sensitivity of <0.006% (range 0.003%−0.03%). The median MFC-‐‑
MRD positive result was 0.08% (range 0.003%−0.9%). The median number of events was 756 437 (range 200 000-‐‑2 283 554). So far samples from 10 patients in sCR/MFC-‐‑MRD negative have been analyzed by ASO-‐‑PCR and 7/10 were also PCR-‐‑MRD negative with a median sensitivity of <0.0006% (range <0.0004%−0.02%). There was no difference between study arms with regard to PFS during the very short-‐‑term follow-‐‑up (Figure 15).
Table 31. Response rates before and after mobilization by intention to treat, N=80 Response before
mobilization, n (%) Response 3 months after ASCT, n(%)
MFC-MRD negative* 22 (28) 28 (35)
sCR 7 (9) 14 (18)
CR 4 (5) 6 (8)
VGPR 25 (31) 26 (33)
PR 33 (41) 12 (15)
PD 2 (3) 9 (11)
Out 9 (11) 13 (16)
ASCT, autologous stem cell transplantation; MFC, multiparameter flow cytometry; MRD, minimal residual disease; *Immunophenotypic remission independently of paraprotein response. sCR; stringent complete remission; CR, complete remission; VGPR, very good partial remission; PR, partial remission;
PD, progressive disease.
Figure 14. Response rates before mobilization (Pre-Mob) and three months after autologous transplantation (3 months Post-ASCT), N=80
Figure 15. Progression-free survival according to mobilization arm.
5.3.3 Adverse events
Fifty grade 3 or higher severe adverse events were documented in 37 patients during RVD induction (Table 28). The most common were infections, like pneumonia and bronchitis, and neutropenia. Nine patients with early drop out due to toxicities had the following adverse events: severe liver toxicity, severe sepsis syndrome with rash, neutropenia, amaurosis fugax, simultaneous diagnosis of lung adenocarcinoma and previous prostate cancer. Two patients were not eligible for ASCT due to infections and one patient died from hepatorenal syndrome. During maintenance 8/59 (14%) of patients have had lenalidomide dose reduction because of neutropenia (n=7) or rash (n=1). Three (4%) patients have discontinued maintenance due to severe rash.
5.4 DRUG SENSITIVITY AND RESISTANCE TESTING (IV) 5.4.1 Chemosensitivity groups based on ex vivo drug sensitivity
Drug sensitivity and resistance testing (DSRT) was applied to 50 samples from NDMM (n=16) and RRMM (n=27) patients against 308 drugs or molecules. Fourteen samples were from the patients included in the FMG-‐‑MM02 study. An example of a waterfall blot with selective DSRT results for an individual MM patient is shown in Figure 16. Unsupervised clustering of both drug sensitivity score (DSS) and selective DSS for each patient resulted in four different patient groups based on their similar response to the tested drugs: group I, sensitive; group II, moderately sensitive; group III, resistant and group IV, highly resistant (Figure 17). Group I and III samples were further divided into two different subgroups A and B based on dexamethasone sensitivity. The samples with low dexamethasone sensitivity clustered mostly in the B subgroups. The result was not so distinct in group II, and the patients in group IV were not sensitive to dexamethasone.
Figure 16. An example of waterfall blot demonstrating the most effective, selective drugs ex vivo marked as read (ABT-199, BMS-754807) and the least effective, non-selective drugs as blue (paclitaxel, docetaxel) for this newly diagnosed MM patient.
In general, conventional chemotherapeutic drugs and proteasome inhibitors showed similar ex vivo sensitivity to patient cells as the healthy control cells. However, the myeloma cells showed variable responses to targeted agents including many signal transduction inhibitors. In group I, the myeloma cells showed sensitivity to several signaling pathway inhibitors, such as CDK, HDAC, IGF-‐‑1R, MEK, PI3K-‐‑mTOR and HSP90 inhibitors (Figure 17; Figure 1a, IV). Cells in group II responded to many of the same drugs as in group I, but the sensitivities were moderate, in particular to HDAC, MEK, and HSP90 inhibitors.
Samples in group III were resistant to targeted therapies and exhibited an overall diminished response to most drugs compared to healthy controls. The samples in group IV showed resistance to almost all drugs tested, except BCL2 inhibitors and bryostatin 1, Figure 17. Responses to glucocorticoids and IMIDs were heterogeneous and varied across the DSRT groups (Figure 17). Differential sensitivities between the different groups to select targeted agents are shown in Figure 18.
Figure 18. Ex vivo responses to proteasome inhibitors and to some targeted drugs. The p-level indicates significant differences in terms of mean DSS between the following groups:
bortezomib; group IV vs all others; carfilzomib, healthy vs group I and IV, I vs III and IV, IV vs II and III; panobinostat, healthy vs I and IV, I vs II, III and IV, IV vs II and III; dual PI3K-mTOR inhibitor, healthy vs I and II, I vs III and IV, II vs III and IV; trametinib, healthy vs I, I vs II, III and IV; navitoclax, healthy vs I, I vs II, III and IV.
Cytotoxic effects of targeted therapies were selective towards CD138+ cells compared to CD138− non-‐‑plasma cells from the same patient, supporting the hypothesis that novel drugs produce a response by targeting important oncogenic signals activated in myeloma cells (Suppl. Figure S5, IV).
5.4.2 DSRT results based on cytogenetic FISH aberrations
The cytogenetic FISH findings of tested samples are showed in relation to the chemosensitivity groups in Figure 19. Samples of patients with t(4;14) clustered predominantly in group II and highly resistant patients in group III-‐‑IV more often had del (17p). When different karyotypes analysed by FISH were evaluated in terms of drug sensitivity the samples of both t(4;14) and del(17p) patients showed more sensitivity to targeted agents.
Figure 19. Ex vivo drug responses based on cytogenetic aberrations. Red boxes represent the cytogenetic aberration of that patient sample.
Possible beneficial therapies for t(4;14) by ex vivo analysis include proteasome inhibitors (bortezomib and carfilzomib tested), pomalidomide, cyclin-‐‑D kinase-‐‑, dual-‐‑PI3K-‐‑mTOR-‐‑, IGF-‐‑1R-‐‑ and HDAC (e.g panobinostat) inhibitors and tosedostat. For the highest risk patients with del(17p) some new possibilities to investigate in vivo were identified. These cells were selectively highly vulnerable to the pan BCL2-‐‑inhibitor navitoclax (Figure 20) and to a lesser extent the selective BCL2-‐‑inhibitor venetoclax (Figure 3b, 3c, Suppl. Figure S6a, IV). HDAC inhibitors (e.g panobinostat) and prima-‐‑1-‐‑Met were among the drugs to which myeloma cells with del(17p) in general showed some sensitivity.
Figure 20. Drug sensitivities according to cytogenetic aberrations
There has been a hope that FGFR-‐‑inhibitors would offer targeted treatment for t(4;14) patients having overexpression of FGFR3 and MMSET genes, but in this ex vivo assay the FGFR inhibitors dovitinib and NVP-‐‑BGJ398 showed no efficacy toward MM cells. Of note, GSK-‐‑J4, an inhibitor of JMJD3/KDM6B was toxic to the cells harboring t(4;14).
The first individual ex vivo -‐‑ in vivo translations were promising in three patients who were treated based on DSRT results: two patients with pomalidomide + dexamethasone and one patient with temsirolimus + bortezomib (Figure 4a-‐‑4f, IV). Patient R-‐‑MM-‐‑2757 had the highest ex vivo sDSS for temsirolimus with a fast in vivo response (Figure 4d, IV). After the second relapse, a decrease in ex vivo sensitivity to temsirolimus, correlating to the in vivo resistance to the drug (Figure 4f, IV).
Finally, we analyzed the value of DSRT results for the prediction of outcome of these mostly relapsed/refractory patients in this study. Time to the next treatment correlated with the DSRT subgroup so that the patients of group I and IV had the shortest TTNT and OS calculated from the date of sample collection, which was always done at the start of new treatment (Figure 21). Patients in the most sensitive group I by ex vivo testing had progression with very short TTNT, and the shortest OS with a hazard ratio of 4.66 (CI95%
1.71 – 12.77), p=0.03. The DSRT chemosensitivity group was the strongest variable for OS in a multivariate analysis including gender, paraprotein type, high-‐‑risk cytogenetics and general sensitivity based on clinical response. For clinical response variable patients were classified as treated and sensitive, alkylator-‐‑refractory, LEN-‐‑refractory, BZM-‐‑refractory, LEN and BZM-‐‑refractory or alkylator-‐‑ LEN-‐‑ and BZM-‐‑refractory.
Figure 21. Overall survival according to DSRT group
6 Discussion
6.1 THE MAIN FINDINGS
These studies focus on the treatment of MM with novel drugs with special reference to the quality of response, autologous stem cell mobilization and translational research with the goal to identify new molecules for targeted therapy especially in high-‐‑risk MM. Studies I-‐‑II consist of 47 NDMM patients, including 129 simultaneous sequential serum, urine and bone marrow samples from 22 nCR/CR patients for MFC and PCR. Study III included 80 NDMM patients in a randomized mobilization substudy after RVD induction and study IV 43 ND or RRMM patients with 50 bone marrow samples evaluated for drug sensitivity testing.
Twenty-‐‑eight percent of all patients (13/47) and 59% (13/22) of nCR/CR patients achieved MolR after BZM+Dex + ASCT treatment. These patients achieved a significantly longer PFS than patients without MolR (Studies I-‐‑II). With the individually expanded clonal design the applicability of ASO-‐‑PCR was 100%, which enabled the comprehensive comparison of MRD methods. MFC and PCR had concordance in the majority of samples but in 35% of cases PCR showed MRD in MFC-‐‑negative samples. The S-‐‑FLC ratio and IFE were not predictive of MRD-‐‑negativity (Studies I-‐‑II). Stem cell mobilization with G-‐‑CSF alone produced comparable CD34+ cell yields with a comparable need for plerixafor than CY + G-‐‑
CSF after lenalidomide-‐‑based induction, but with more aphereses (Study III).
The DSRT method was applicable in MM cells ex vivo, and four different sensitivity profiles were recognizable. Group I consisted of samples showing sensitivity to molecules targeting different signaling pathways and these patients had shortest TTNT and OS.
Myeloma cells of high-‐‑risk del17p patients were refractory to almost all compounds, but showed the highest selective sensitivity to BCL2 inhibitors (Study IV).
6.2 PATIENTS
The study patients in these clinical studies represent common, non-‐‑selected NDMM patients in the routine hospital daily ward. The proportion of high-‐‑risk patients based both on FISH findings (22% in FMG-‐‑MM01 and 20% in FMG-‐‑MM02), correlates also with the earlier published data in MM population (51,54). During the course of these studies in 2008 – 2016 Boyd et al. (348) and thereafter IMWG (86) published guidelines for risk stratification in MM both based on ISS and cytogenetics and these were again revised by Palumbo et al.
in 2015 (85). This indicates that along with the development and implementation of novel drugs the sorting of patients by parameters predicting outcome has recently been one of the main interests. The patients in these studies have been classified based on IMWG ISS criteria (84) with the majority of patients included in stage II.
During the last decade it has become clear that novel drugs benefit most the patients with standard-‐‑risk multiple myeloma, but patients with high-‐‑risk and relapsed, refractory myeloma still have a poor outcome. Development of new treatment strategies for these patients is an important challenge. Because the aim of the FMG-‐‑MM01 study was to explore the MolR rate in nCR/CR patients the patients whose response was worse than PR after two BZM+Dex cycles were excluded from this study, so the results obtained in PFS and OS demonstrate the outcome of early well-‐‑responding patients. However, when OS between patients off-‐‑ and on-‐‑ protocol were compared later on there was no statistical difference in the long-‐‑term follow-‐‑up suggesting the efficacy of second line and later salvage treatments.
The proportion of HR patients in early off-‐‑protocol group was 59% and 30% in the group continued per protocol. In the off-‐‑protocol group then 41% were standard-‐‑ and low-‐‑risk
patients not achieving PR after two BZM+Dex cycles. They possibly had an MGUS-‐‑like MM profile, which could explain the similar OS in the long-‐‑term follow-‐‑up between off-‐‑ and on-‐‑
protocol group. The negative impact of high-‐‑risk cytogenetics on PFS was clearly seen also in this small MM population. On the other hand, patients with better outcome could be distinguished also based on conventional CR response, in line with previous publications (278). Still, within the CR group the molecular response can give additional information dividing the patients in two separate groups with different outcomes (Figure 4, II).
In the DSRT study the samples were collected from 16 (37%) randomly selected NDMM patients and 27 (63%) RRMM patients. This study focused first on very refractory patients to evaluate the applicability of the assay in patients with immediate need for help. After confirmation of the applicability the study was expanded to NDMM patients.
6.3 RESPONSE ASSESSMENT METHODS (I, II)
The depth of treatment response is a direct surrogate for outcome in MM (349). Beside conventional serological response markers the assessment of the deeper quantitative response calls for a practical, economic, reliable and standardized method to detect MRD.
Both MFC and PCR have been investigated in several studies (Table 16, Table 17) and both have advantages and disadvantages (Table 19). In practice MFC with faster analysis has awoken wider interest than the more time-‐‑consuming and laborious techniques of PCR.
More patients achieve CR/sCR with novel treatments even in relapsed disease setting (214), increasing pressure to develope of high-‐‑quality MRD assays to compare the efficacy of different expensive treatment regimens increases.
The challenge in PCR is the design of a suitable probe. This has been successful in 42-‐‑
86% of cases in earlier publications (Table 17) and is usually at the level of 75% compared to at least 95% of detecting the aberrant PC clone by MFC (Table 16). In our study the PCR coverage of 100% could be achieved by gradual widening of the primer sets used for the clonality detection. Somatic hypermutations characteristic of MM can complicate the detection of a suitable clonal mutation and design of the probe. Using only a standardized method for ASO-‐‑PCR with only one specific forward primer can lead to low PCR applicability (318). The detection of the clone may be improved by sequencing of Ig light chain, IgH and kappa-‐‑deleting elements (Kde) rearrangements and alternate locations for consensus primers (310, I and II). In our study the 100% applicability was reached in unselected nCR/CR patients by using singleplex Ig consensus primers in addition to the multiplex primers, inclusion of the light chain rearrangements for clone detection and if needed, using a reverse-‐‑oriented ASO-‐‑primer and individually designed TaqMan probe for ASO-‐‑PCR analysis.
This enhanced three-‐‑step ASO-‐‑PCR approach will increase the costs of assay by 30% and could have the most benefit in multicenter trials for detecting MRD in CR patients. Due to the need for a fast and practical method for routine MRD assay the sensitivity of MFC could be improved by using 8−10 or even 12 color MFC and trying to collect up to 2−5 x 106 cells using a bulk lysis method (13, 14, 92, 93, 94). This will prolong the time needed for flow cytometry run and probably will require additional devices, which will also increase the costs of MFC. However, the sensitivity of MFC can never reach that of PCR because MFC requires 10−50 clonal cells to confirm MRD, but an optimal molecular method can detect even a single clonal residual cell, which gives at least one logarithm higher sensitivity to molecular methods in any case.
MRD is usually measured in patients when a nCR/CR response has been achieved corresponding to serum paraprotein concentration of 150–500 mg/l (350). However, because of the long half-‐‑life of paraprotein a patient with IFE positivity may be in MolR. Persisting IFE positivity may, on the other hand, represent oligoclonal secretion called atypical serum IFE pattern (ASIP), i.e. appearance of either heavy or light chain components which differ from the original paraprotein. In one study up to 33% of autotransplanted patients
developed ASIP and 4/7 analyzed patients were in MolR (351). Occurrence of ASIP has been fairly common (up to 60%) after novel therapies compared to conventional therapies (11.1%) (352). However, the sustained primary IFE positivity in MFC-‐‑MRD negative patients predicts a shorter PFS (285).
The FLC assay can detect light chains below the concentration of 1 mg/l. However, an aberrant FLC ratio can be produced also by oligoclonal secretion. In the study by Singhal et
The FLC assay can detect light chains below the concentration of 1 mg/l. However, an aberrant FLC ratio can be produced also by oligoclonal secretion. In the study by Singhal et