Salvage therapy versus upfront autologous stem cell transplantation in multiple myeloma patients with progressive disease after first-line induction therapy

Abstract

It is a matter of debate whether myeloma patients with progressive disease (PD) after induction should receive salvage therapy or proceed directly to autologous stem cell transplantation. We performed a retrospective analysis of 1599 patients treated between 1991 and 2016 at the University Hospital of Heidelberg and other centers. Deepening of response through salvage therapy did not lead to better progression-free or overall survival (PD versus salvage therapy patients: HR = 0.71, 95% CI [0.28, 1.80], p = 0.5 and HR = 0.77, 95% CI [0.30, 1.95], p = 0.6, respectively), neither in patients treated with novel agents (HR = 0.66, 95% CI [0.23, 1.85], p = 0.4 and HR = 0.76, 95% CI [0.27, 2.15], p = 0.6) nor older regimens (HR = 0.86, 95% CI [0.36, 2.07], p = 0.7 and HR = 0.8, 95% CI [0.34, 1.91], p = 0.6). Therefore, primary nonresponders might benefit from a direct transplant rather than salvage induction, although the analyzed salvage therapy cohort was small (n = 23) and cytogenetics was not included in the multivariable analysis.

Keywords:

• Multiple myeloma
• salvage therapy
• autologous stem cell transplantation

Introduction

High dose therapy (HDT) with melphalan, followed by autologous stem cell transplantation (ASCT) was introduced as an effective treatment for patients with multiple myeloma (MM) more than 30 years ago [1,2]. Depth of response before HDT/ASCT is of prognostic significance for overall and progression-free survival (OS and PFS, respectively) [3]. However, it is unclear whether nonresponders to induction therapy should proceed directly to ASCT or undergo a salvage therapy to deepen their response. Regarding the partly contradictory results of previous reports on this subject [4–10], we conducted a retrospective analysis of a large cohort of newly diagnosed MM patients who had received HDT/ASCT as primary therapeutic concept.

We analyzed the prognostic significance of response, in particular of progressive disease (PD) compared to at least stable disease (SD) before first ASCT as well as performing of salvage therapy in initial nonresponders to induction therapy. Because of the large time span between the first and last included patients, we additionally performed subgroup analyses of patients who received novel agents in induction therapy and those treated with older regimens.

To date, there are no known factors associated with occurrence of PD before ASCT. We could show a relation of clinical/therapeutic factors associated with PD after induction therapy in all analyzed patients as well as those treated with and without novel agents in induction.

To reflect the real-world setting of the analysis, both nontrial and clinical trial patients treated with different induction regimens, single or tandem ASCT, and different or no maintenance therapy were evaluated.

Patients and methods

Patients

We included 1599 patients who had undergone ASCT as primary therapy for MM in 1991–2016. Of these, 931 (58.2%) were nontrial patients (NTP), whereas 344 (21.5%) were treated within the German-Speaking Myeloma Multicenter Group (GMMG)-HD3 [11] [comparison of vincristine, doxorubicin, and dexamethasone (VAD) in induction and interferon in maintenance therapy with thalidomide, doxorubicin, and dexamethasone (TAD) in induction and thalidomide in maintenance therapy] and 324 (20.3%) in the GMMG-HD4 [12] (induction with VAD and thalidomide maintenance therapy versus bortezomib, doxorubicin, and dexamethasone (PAD) in induction and bortezomib maintenance therapy) trial. All NTP were transplanted at the Heidelberg University Hospital, whereas trial patients underwent ASCT either in Heidelberg or other GMMG centers. Performing of the ASCT in NTP was an individual decision based on clinical eligibility of patients and their age (usually up to 70 years, but the biological rather than metric age was the crucial criterium). Tandem ASCT was performed in patients who had received less than complete remission (CR) after the first ASCT and were clinically eligible for a second transplant. Novel agents in induction therapy were used in 798 (49.9%) individuals. Regarding the induction regimens, 725 (45.3%) patients received VAD; 379 (23.7%) PAD; 212 (13.3%) thalidomide, doxorubicin, and dexamethasone (TAD); 143 (8.9%) bortezomib, cyclophosphamide, and dexamethasone (VCD), and 146 (9.1%) other regimens, including agents such as lenalidomide, thalidomide, bortezomib, cyclophosphamide, etoposide, idarubicin, cisplatin, doxorubicin, melphalan, rituximab, bendamustine, dexamethasone, and prednisone. A subgroup of patients (n = 23, 1.4%) received salvage therapy due to PD and thereby deepened their response before first ASCT.

Salvage therapy was administered to patients with PD after the induction therapy who were clinically fit enough and willing to undergo such treatment in order to improve the response depth before ASCT. Of these patients, 12 (52.2%) received and 11 (47.8%) did not receive novel agents in induction therapy. In the novel agents group, 4 (33.3%) patients received thalidomide, cyclophosphamide, etoposide, and dexamethasone (TCED) and VAD; 2 (16.7%) patients lenalidomide and dexamethasone (RD) and VCD; 2 (16.7%) patients TCED and vincristine, idarubicin, and dexamethasone (VID); 1 (8.3%) patient VAD and VCD; 1 (8.3%) patient VID, dose-reduced cyclophosphamide, doxorubicin, vincristine, and prednisone (mini-CHOP), ifosfamide, and thalidomide, cyclophosphamide, vincristine, and dexamethasone (TCVD); 1 (8.3%) patient melphalan and prednisone (MP), TCED, and VAD; and 1 (8.3%) patient TCED, VAD, and VID. Among the salvage patients who did not receive novel agents in the induction therapy, 4 (36.4%) patients received cyclophosphamide, etoposide, and dexamethasone (CED) and VAD; 3 (27.3%) patients CED and VID; 2 (18.2%) patients MP and VAD; 1 (9.1%) patient MP and VID; and 1 (9.1%) patient MP, CED, and VAD.

In total, 820 (51.3%) patients underwent single and 779 (48.7%) tandem ASCT. Maintenance therapy was administered to 921 (62.0%) patients. Among these, 367 (39.9% of maintenance patients) received thalidomide, 362 (39.3%) interferon, 160 (17.4%) bortezomib, 22 (2.4%) lenalidomide, and 27 (2.9%) other maintenance therapy, such as ixazomib, rituximab, and lenalidomide/dexamethasone. In total, 542 individuals (59.1% of maintenance patients) received novel agents during maintenance therapy. Median follow-up was 83 months since the first ASCT. During the follow up, 775 deaths and 1227 PFS events were observed. Clinical factors were evaluated at induction therapy start and are presented in Table 1. The median age was 57 years and the same in trial and NTP. Cytogenetic data, evaluated by means of interphase fluorescent in situ hybridization, was evaluable in 691 (43.2%) patients. High-risk cytogenetics were defined as the presence of at least one of the following: >1q21 gain (>3 copies) (n = 256, 36.9%), del(17p13) (n = 62, 8.8%), or t(4;14) (n = 87, 12.4%). In total, 320 (46.3%) patients had high-risk cytogenetics. The International Staging System (ISS) was evaluable in 1038 (65.9%) patients. ISS I was present in 457 (44.0%), ISS II in 336 (32.4%), and ISS III in 245 (23.6%) individuals. Disease status was assessed according to the criteria of the European Society for Blood and Marrow Transplantation (EBMT) [13], the standard tool for response evaluation during the treatment of most patients included.

Table 1. Patients’ characteristics.

	All (n = 1559)		PD (n = 120)		Salvage (n = 23)
	n	% all	n	% PD	n	% salvage
Cohort
HD3	344	21.5	47	39.2	0	0
HD4	324	20.3	16	13.3	0	0
NTP	932	58.2	57	47.5	23	100
All	1599	100	120	100	23	100
Sex
Male	934	58.4	66	55.0	14	60.9
Female	665	41.6	54	45.0	9	39.1
All	1599	100	120	100	23	100
Age
≤60 years	1042	65.2	80	66.7	18	78.3
>60 years	557	34.8	40	33.3	5	21.7
All	1599	100	120	100	23	100
LDH
Normal	992	84	81	85.3	4	100
High	189	16	14	14.7	0	0
All	1181	100	95	100	4	100
ISS
I	457	44	26	31.7	1	50
II	336	32.4	30	36.6	1	50
III	245	23.6	26	31.7	0	0
All	1038	100	82	100	2	100
High-risk cytogenetics
No	371	53.7	22	50	1	100
Yes	320	46.3	22	50	0	0
All	691	100	44	100	1	100
NA in IT
No	801	50.1	68	56.7	11	47.8
Yes	798	49.9	52	43.3	12	52.2
All	1599	100	120	100	23	100
ASCT
Single	820	51.3	60	50	18	78.3
Tandem	779	48.7	60	50	5	21.7
All	1599	100	120	100	23	100
Maintenance
No	565	38	43	37.1	2	11.8
Yes	921	62	73	62.9	15	88.2
All	1486	100	116	100	17	100
NA in maintenance
No	375	40.9	32	45.1	9	75
Yes	542	59.1	39	54.9	3	25
All	917	100	71	100	12	100
Creatinine
≤2 mg/dl	1130	89.6	88	88.9	2	50
>2 mg/dl	131	10.4	11	11.1	2	50
All	1261	100	99	100	4	100
Hemoglobin
≥10 g/dl	867	67.7	63	63	1	25
<10 g/dl	413	32.3	37	37	3	75
All	1280	100	100	100	4	100
Platelets
≥150/nl	1130	89.8	91	90.1	3	75
<150/nl	128	10.2	10	9.9	1	25
All	1258	100	101	100	4	100
Calcium
≤2.75 mmol/l	1139	92	94	94	2	50
>2.75 mmol/l	99	8	6	6	2	50
All	1238	100	100	100	4	100
Bone lesions
None	268	17.8	24	21.1	6	28.6
1–2	157	10.4	12	10.5	3	14.3
>2	1083	71.8	78	68.4	12	57.1
All	1508	100	114	100	21	100
Response before first ASCT
CR	85	5.3
PR	1148	71.8
MR	155	9.7
SD	91	5.7
PD	120	7.5
All	1599	100
Response after first ASCT
CR	169	10.8	0	0	0	0
PR	1211	77.5	63	55.3	16	76.2
MR	101	6.5	33	28.9	2	9.5
SD	35	2.2	13	11.4	1	4.8
PD	46	2.9	5	4.4	2	9.5
All	1562	100	114	100	21	100

ASCT: autologous stem cell transplantation; IT: induction therapy; n: number of patients; NTP: nontrial patients; NA: novel agents; PD: progressive disease.

Methods

Patients with response assessment prior to ASCT were analyzed. PFS and OS were calculated from the date of first ASCT. Survival times were estimated using the Kaplan–Meier method and compared with the log-rank test. Multivariable Cox and logistic regression models were used to assess the prognostic impact of clinical factors. Response after 1st ASCT, 2nd ASCT, and maintenance therapy start were analyzed as time-dependent factors in Cox regression. In the multivariable Cox and logistic regression models, missing values of baseline parameters were imputed using the MICE algorithm [14] based on B = 50 bootstrap runs. Multivariable models always included age >60 years, ISS II and III, novel agents in induction therapy, patient cohort, response before and after 1st ASCT, tandem ASCT, and maintenance therapy plus additional factors depending on subgroup and objective such as novel agents in maintenance therapy, high-risk cytogenetics, creatinine >2 mg/dl, calcium >2.75 mmol/l, hemoglobin <10 g/dl, platelets <150/nl, and bone lesions. p-values are two-sided. p-values below 0.05 were considered statistically significant. Analyses were performed using statistical software R 3.4 [15].

Results

Prognostic significance of response before 1st ASCT

Of all 1599 analyzed patients, 85 (5.3%) achieved CR, 1148 (71.8%) a partial remission (PR), 155 (9.7%) a minimal response (MR), and 91 (5.7%) SD before first ASCT. A subgroup of patients (n = 120, 7.5%) progressed between the last induction course and ASCT.

We compared the PFS and OS of patients with PD (n = 120) to those with at least SD (= non-PD, n = 1479) before first ASCT. The univariable analysis did not show a statistically significant difference in PFS or OS of PD compared to non-PD patients (22.9 vs. 27.4 months, p = 0.08 and 69.5 vs. 86.5 months, p = 0.1, respectively) (Figure 1(A,B)). Similarly, there was no significant difference in PFS and OS between the PD and non-PD patients on multivariable analysis (hazard ratio (HR) = 1.23, 95% confidence interval (CI) [0.98, 1.56], p = 0.08 and HR = 1.24, 95% CI [0.93, 1.65], p = 0.1, respectively).

Figure 1. Prognostic significance of response and salvage therapy before first ASCT: all induction regimens. Survival of patients with non-PD versus PD (A, B) before first ASCT as well as salvage versus PD patients (C, D). ASCT: autologous stem cell transplantation; non-PD: responders with ≥ stable disease to the first induction or salvage therapy; OS: overall survival; PD: progressive disease; salvage, patients with ≥ stable disease due to salvage therapy; PFS: progression-free survival.

We further performed subgroup analyses to see if prognostic significance of response before first ASCT differed in patients who were treated with novel agents during induction from those who were not.

Among the group receiving novel agents, PD patients (n = 52) had significantly worse PFS and OS compared to non-PD patients (n = 746) (22.2 versus 29.1 months, p = 0.03 and 54.4 versus 97.5 months, p < 0.001, respectively) on univariable analysis (Figure 2(A,B)). The multivariable analysis confirmed these findings with worse PFS and OS of PD compared to non-PD patients (HR = 1.58, 95% CI [1.09, 2.30], p = 0.02 and HR = 2.06, 95% CI [1.30, 3.27], p = 0.002, respectively). This might be explained by a prognostically disadvantageous disease biology in patients nonresponsive to novel agents.

Figure 2. Prognostic significance of response and salvage therapy before first ASCT: novel agents in induction/salvage. Survival of patients with non-PD versus PD (A, B) before first ASCT as well as salvage versus PD patients (C, D): a subgroup analysis of patients treated with novel agents in induction therapy. ASCT: autologous stem cell transplantation; non-PD: responders with ≥ stable disease to the first induction or salvage therapy. PD: progressive disease; salvage, patients with ≥ stable disease due to salvage therapy; PFS: progression-free survival; OS: overall survival.

In contrast to the novel agent group, PD (n = 68) and non-PD patients (n = 733) without novel agents in induction therapy had similar PFS (22.9 versus 26.8, p = 0.7) and OS (84.6 versus 81, p = 0.6) on univariable analysis (Figure 3(A,B)). In line with these findings, the multivariable analysis showed no significant difference in PFS and OS between PD and non-PD patients (HR = 0.98, 95% CI [0.72, 1.34], p = 0.9 and HR = 0.95, 95% CI [0.64, 1.41], p = 0.8, respectively).

Figure 3. Prognostic significance of response and salvage therapy before first ASCT: no novel agents in induction/salvage. Survival of patients with non-PD versus PD (A, B) before first ASCT as well as salvage versus PD patients (C, D): a subgroup analysis of patients who did not receive novel agents in induction therapy. ASCT: autologous stem cell transplantation; Non-PD: responders with ≥ stable disease to the first induction or salvage therapy; OS: overall survival; PD: progressive disease; salvage, patients with ≥ stable disease due to salvage therapy; PFS: progression-free survival.

Effect of salvage therapy before 1st ASCT

We further compared patients who underwent ASCT having PD (n = 120) to those who had received salvage therapy and improved their response to at least SD before ASCT (“salvage patients”, n = 23).

In contrast to the all non-PD versus PD patients, the univariable analysis showed worse PFS (12.1 versus 22.9 months, p = 0.04) and OS (33.1 versus 69.5 months, p = 0.08) of patients who received salvage therapy compared to those with PD before ASCT (Figure 1(C,D)). On multivariable analysis, no significant difference between the PD and salvage group was observed for PFS and OS (HR = 0.71, 95% CI [0.28, 1.80], p = 0.5 and HR = 0.77, 95% CI [0.30, 1.95], p = 0.6, respectively).

Similarly, among patients who received novel agents in induction therapy, salvage patients (n = 12) also had significantly worse PFS and OS compared to PD patients (n = 52) (10.6 versus 22.2 months, p = 0.001 and 21 versus 54 months, p = 0.005) on univariable analysis (Figure 2(C,D)), while the multivariable analysis showed no significant difference in PFS and OS between PD and salvage patients (HR = 0.66, 95% CI [0.23, 1.85], p = 0.4 and HR = 0.76, 95% CI [0.27, 2.15], p = 0.6, respectively).

Among patients who did not receive novel agents in induction therapy, salvage (n = 11) and PD patients (n = 68) showed no significant difference in PFS (45.4 versus 22.9 months, p = 0.7) and OS (71.3 versus 84.6 months, p = 0.3) on univariable analysis (Figure 3(C,D)). In line with these results, the multivariable analysis did not reveal a significant difference in PFS and OS between the PD and salvage group (HR = 0.86, 95% CI [0.36, 2.07], p = 0.7 and HR = 0.8, 95% CI [0.34, 1.91], p = 0.6, respectively).

Prognostic factors in patients with PD before 1st ASCT

When analyzing all patients regardless of the induction regimen, the multivariable analysis showed an association between age >60 years and worse PFS (HR = 2.47, 95% CI [1.42, 4.27], p = 0.001) as well as of tandem ASCT (HR = 0.51, 95% CI [0.31, 0.85], p = 0.01) and a borderline association of maintenance therapy (HR = 0.58, 95% CI [0.34, 1.00], p = 0.05) with better PFS (Supplementary Table S1). Maintenance therapy was a significant prognostic factor for better OS in PD patients (HR = 0.45, 95% CI [0.23, 0.88], p = 0.02), while none of the factors were related to worse OS (Supplementary Table S2).

Among PD patients with novel agents in induction, only ISS II was a prognostic factor for worse PFS (HR = 0.29, 95% CI [0.09, 0.95], p = 0.04), whereas tandem ASCT was for better PFS (HR = 0.24, 95% CI [0.09, 0.64], p = 0.004) (Supplementary Table S1). No factors affected OS to a statistically significant extent (Supplementary Table S2).

Among PD patients who did not receive novel agents during induction therapy, only age >60 years was a significant prognostic factor for worse PFS (HR = 4.06, 95% CI [1.83, 8.97], p = 0.0006) (Supplementary Table S1), while no factors affected the OS in a significant way (Supplementary Table S2).

Furthermore, as poor/nonresponders to an induction therapy with bortezomib have been shown to have worse PFS compared to good responders [16], we examined whether poor responders (MR/SD/PD) and/or progressive patients after an induction with bortezomib presented worse PFS/OS compared to poor responders/progressive patients who had received other induction treatments. Among the poor responders (PD/SD/MR) (n = 306), there was no statistical difference between patients treated with (n = 76) and those not treated with bortezomib (n = 230) in PFS (23.1 and 24.6 months, p = 0.6) and OS (not reached (NR) and 77.8 months, p = 0.3) on univariable analysis (Supplementary Figure S1). The multivariable analysis confirmed these results with no significant difference in PFS and OS between poor responders to bortezomib and other agents (HR = 1.02, 95% CI [0.73, 1.43], p = 0.9 and HR = 0.76, 95% CI [0.47, 1.22], p = 0.3, respectively). When analyzing only patients with PD after induction therapy with bortezomib (n = 25) to PD patients after induction with other agents (n = 68), the bortezomib-treated group did not show a significant difference in PFS or OS compared to the nonbortezomib group in PFS (22.2 and 22.9 months, p = 0.4) and OS (62.1 and 84.6 months, p = 0.5) on univariable analysis (Supplementary Figure S2). Similarly, the multivariable analysis did not show a significant difference between the progressive patients treated with compared to without bortezomib during induction therapy in PFS or OS (HR = 1.28, 95% CI [0.64, 2.56], p = 0.5 and HR = 1.25, 95% CI [0.52, 2.98], p = 0.6, respectively).

Prognostic factors for PD before first ASCT

Among all analyzed patients, prognostic factors associated with occurrence of PD before first ASCT on multivariable analysis were ISS II and III (odds ratio (OR) = 1.75, 95% CI [1.26, 3.31], p = 0.02 and OR = 2.83, 95% CI [1.69, 5.44], p = 0.0005, respectively) as well as treatment within the GMMG-HD4 trial and the nontrial group (OR = 0.3, 95% CI [0.16, 0.54], p = 0.00008 and OR = 0.39, 95% CI [0.26, 0.61], p = 0.00001, respectively).

Patients treated with novel agents during induction therapy more often had PD before first ASCT when treated within the GMMG-HD4 trial (OR = 0.1, 95% CI [0.02, 0.36], p = 0.002).

Among patients treated without novel agents in induction, ISS III (OR = 3.1, 95% CI [1.87, 8.39], p = 0.003) as well as treatment within the GMMG-HD4 trial and the nontrial group (OR = 0.4, 95% CI [0.18, 0.76], p = 0.01 and OR = 0.26, 95% CI [0.13, 0.45], p = 0.00001, respectively) were associated with PD before first ASCT (Table 2).

Table 2. Prognostic factors for progressive disease before first ASCT. Multivariable logistic regression model.

	All (n = 1599)			With NA in IT (n = 798)			Without NA in IT (n = 801)
	OR	95 % CI	p	OR	95 % CI	p	OR	95 % CI	p
Age >60 years	0.88	[0.57, 1.30]	0.6	0.95	[0.49, 1.66]	0.9	0.82	[0.45, 1.41]	1
ISS II	1.75	[1.26, 3.31]	0.02	1.37	[0.53, 2.49]	0.4	1.87	[0.86, 3.40]	0.08
ISS III	2.83	[1.69, 5.44]	0.0005	1.95	[1.15, 6.01]	0.1	3.1	[1.87, 8.39]	0.003
High LDH	0.84	[0.38, 1.22]	0.6	1.13	[0.59, 2.60]	0.7	0.6	[0.19, 1.07]	0.2
NA in IT	0.71	[0.48, 10.3]	0.08
Creatinine >2 mg/dl	0.88	[0.49, 1.82]	0.7	1.16	[0.28, 2.02]	0.8	0.68	[0.29, 1.84]	0.4
Hemoglobin <10 g/dl	1.02	[0.68, 1.65]	1	0.76	[0.36, 1.49]	0.4	1.30	[0.77, 2.45]	0.4
Calcium >2.75 mmol/l	0.58	[0.20, 10.7]	0.2	0.33	[0.04, 0.98]	0.1	0.83	[0.30, 1.97]	0.7
Platelets <150/nl	0.74	[0.42, 1.54]	0.4	0.79	[0.28, 2.22]	0.7	0.74	[0.40, 2.10]	0.5
HD4	0.3	[0.16, 0.54]	0.00008	0.1	[0.02, 0.36]	0.002	0.4	[0.18, 0.76]	0.01
NTP	0.39	[0.26, 0.61]	0.00001	0.57	[0.29, 1.05]	0.08	0.26	[0.13, 0.45]	0.00001
1–2 bone lesions	0.9	[0.45, 1.89]	0.8	0.81	[0.21, 2.62]	0.7	0.94	[0.36, 2.44]	0.9
>2 bone lesions	0.72	[0.49, 1.30]	0.2	0.88	[0.45, 2.14]	0.7	0.59	[0.36, 1.34]	0.1

CI: confidence interval; IT: induction therapy; OR: odds ratio; NA: novel agents; NTP: nontrial patients; n: number of patients.

Discussion

HDT with melphalan, followed by ASCT has been shown to be the most effective primary therapy for MM patients who are well and young enough to undergo such treatment [1,2,17–19]. Whether patients with PD should be given salvage therapy in order to deepen their response before first ASCT, is a matter of debate.

The analysis of all PD (PD after the first induction + PD after salvage therapy) compared to all non-PD (sensitive to first induction + sensitive to salvage therapy) patients showed no significant difference in PFS or OS of both response groups. The same results were seen when performing a subgroup analysis of patients who did not receive novel agents during induction therapy. However, among patients treated with novel agents in the induction, patients with PD before first ASCT had significantly worse PFS and OS compared to the non-PD patients. This indicates that good responders to induction therapy, who were treated with novel agents have better survival than other response/treatment groups, while patients refractory to induction with novel agents possibly have an adverse disease biology, which is not surprising.

According to our analysis, there was no survival benefit of deepening of the response through salvage therapy compared to ASCT in PD. This was the case independent of whether novel agents had been included in the induction therapy or not. The even worse PFS and OS of all salvage patients compared to all PD patients as well as in the subgroup receiving novel agents in induction on univariable analysis might be due to a cumulative toxic effect of salvage therapy and suggests that even in the era of novel agents, direct proceeding to ASCT is still an effective therapy for transplant-eligible myeloma patients, regardless of their response to induction. An alternative explanation could be that the patients who were offered salvage therapy might have had more aggressive disease than those who did not undergo salvage therapy. These, however, remain speculative statements and our data does not provide sufficient evidence allowing further conclusions in this regard.

As it is still unclear why some patients do not respond to induction therapy while others do, we looked for prognostic factors associated with PD before first ASCT. When analyzing all patients (with and without novel agents in induction therapy), patients with ISS ≥ II and those treated within the GMMG-HD4-trial and non-trial patients progressed more frequently before first ASCT. Among patients who did not receive novel agents in induction, there was an association between ISS III, treatment in the GMMG-HD4 trial as well as outside of clinical trials and the occurrence of PD before first ASCT. Patients treated with novel agents in induction only had a significantly higher risk of developing PD before first ASCT when treated in the GMMG-HD4 trial. To the authors, the superiority of induction treatment within the GMMG-HD3 over GMMG-HD4 and nontrial patients is unclear at the moment.

In the 1990s, Alexanian et al. [4] stated that among newly diagnosed MM patients, those with primary refractory MM or SD benefit most from ASCT, under the condition that the myeloablative therapy is performed within 1 year after induction start [4]. Singhal et al. [5] and Kumar et al. [6] reported similar event-free survival (EFS) [5], PFS [6], and OS [5,6] between different response groups before ASCT.

On the other hand, in the ad hoc analysis of the Intergroupe Francophone Du Myelome 2005-01 trial, the major prognostic factor negatively affecting PFS was achievement of less than very good partial remission after induction therapy [7]. Two other studies showed poorer PFS and OS in patients with PD compared to those who achieved SD [8] or MR/SD [9] before ASCT. Nevertheless, the results indicated that even patients with PD benefit from ASCT by deepening of the response, as disease status at day 100 after ASCT was a significant prognostic factor for OS, PFS, and relapse rate [9]. Vij et al. described significantly worse PFS and OS in patients with < PR compared to those who achieved PR/CR after first-line induction therapy. At the same time, they suggested patients achieving < PR should directly undergo ASCT rather than receive salvage therapy, as treatment-related mortality, relapse-free survival, PFS, and OS of patients with < PR after first-line induction therapy and those who achieved < PR after first-line induction but deepened their response after salvage therapy were similar [10]. Recently, the preliminary results from the prospective phase III Myeloma XI Study showed a prolongation of PFS in newly diagnosed MM patients by sequential treatment with cyclophosphamide, bortezomib, and dexamethasone (CVD) after a suboptimal (≤PR) response to 1^st line therapy with an immunomodulatory drug (IMiD), cyclophosphamide, and dexamethasone. Although both transplant eligible (TE) and ineligible patients were enrolled in the trial, the prolonged PFS was particularly significant in the TE cohort [20]. This statement does not contradict our results, as our analysis specifically focused on patients with PD before first ASCT and the subgroup analysis of PD patients in Myeloma XI Study is not available to date. Furthermore, the most significant response upgrade was seen in the MR/PR group [20]. In contrast to the above-mentioned study, we included only TE patients who had received different induction regimens, either with or without novel agents. Another report from the Myeloma XI Study [21] showed a significant prolongation of PFS in initial non-responders to IMiD induction who had received a subsequent therapy with CVD and responded with ≥ MR compared to < MR (19 versus 7 months, respectively). More than half (32/50) of TE patients who responded well to CVD received ASCT and showed a PFS of 27 months. This report, however, focuses on a specific group of patients (refractory to IMiDs, sensitive to CVD), whereas our analysis includes patients with different types of first line and salvage therapy. Furthermore, the nonresponders in the above-mentioned report were defined as patients with a response < MR, while our study focuses on patients with PD compared to ≥ SD before 1st ASCT.

Due to the large time span between the first and last included patients, we were able to provide long follow-up data. However, some of the therapy regimens are currently out of standard use. For that reason, we performed subgroup analyses of patients who received novel agents in induction and those who did not.

Moreover, it needs to be emphasized that, in some groups, the numbers of patients on which our conclusions are based are small (23 patients treated with salvage therapy, of whom 12 did and 11 did not receive novel agents as part of induction therapy). This is reflected by the wide confidence intervals of our findings. Therefore, the power of this study might not be sufficient to draw ultimate conclusions and the results should be interpreted with caution.

According to our results, there was no survival difference between patients with PD and ≥ SD before the first ACST. Considering this finding, the question arises, whether induction therapy is of any use at all. This was not a focus of our analysis, as we only compared the nonresponders to patients who responded with ≥ SD to the induction therapy. Although there is a general tendency to achieve a deep therapy response before ASCT [22], further analyses should be performed to answer this question.

Cytogenetics data, especially of patients treated in the 1990s and early 2000s, was often not evaluated. We therefore could not include it in most of our multivariable analyses, which needs to be taken into account while drawing conclusions of our results as high-/ultra-high risk cytogenetics have shown to have a negative effect on patients’ survival [23].

Moreover, this analysis only includes progressive patients who underwent ASCT. The survival of patients with PD after induction who, for different reasons, had not undergone ASCT, was not assessed.

Disclosure statement

Joanna Blocka and Thomas Hielscher have no relevant conflicts of interest to disclose. Carsten Mueller-Tidow has served on advisory boards with Janssen. Hartmut Goldschmidt received research support to the institution from Amgen, Bristol-Myers Squibb, Celgene, Chugai, Janssen, Mundipharma, Novartis, Sanofi, and Takeda, has served on advisory boards with Adaptive Biotechnology, Amgen, Celgene, Janssen, Sanofi, and Takeda, and received honoraria from ArtTempi, Bristol-Myers Squibb, Celgene, Chugai, Janssen, and Novartis. Jens Hillengass has served as a consultant to Amgen and Celgene, received honoraria from Amgen, Bristol-Myers Squibb, Celgene, Janssen, Novartis, and Takeda, received research funding from Celgene and Sanofi as well as has served on advisory boards with Amgen, Bristol-Myers Squibb, Celgene, Janssen, and Novartis.