COMPARABLE OUTCOMES IN NON-SECRETORY AND SECRETORY MULTIPLE MYELOMA AFTER AUTOLOGOUS STEM CELL TRANSPLANTATION

Abstract

Non-secretory myeloma (NSM) accounts for <5% of cases of multiple myeloma (MM). The outcome of these patients following autologous stem cell transplantation (ASCT) has not been evaluated in clinical trials. We compared the outcomes after ASCT for patients with NSM reported to the CIBMTR between 1989 and 2003, to a matched group of 438 patients (4 controls for each patient) with secretory myeloma (SM). The patients were matched using propensity scores calculated using age, Durie-Salmon stage, sensitivity to pre-transplant therapy, time from diagnosis to transplant and year of transplant. Disease characteristics were similar in both groups at diagnosis and at transplant except higher risk of anemia, hypoalbuminemia and marrow plasmacytosis (in SM) and plasmacytoma (more in NSM). Cumulative incidence of TRM, relapse, PFS and OS were similar between the groups. In multivariate analysis, based on a Cox model stratified on matched pairs and adjusted for covariates not considered in the propensity score, we found no difference in outcome between the NSM and SM groups. In this large cohort of patients undergoing ASCT, we found no difference in outcomes of patients with NSM compared to those with SM.

INTRODUCTION

Non-secretory multiple myeloma (NSM), characterized by no detectable monoclonal protein in the serum or urine by conventional techniques comprises less than 5% of all patients with multiple myeloma.(1–7) The incidence of this condition varies considerably in different series reflecting referral bias as well as the sensitivity of laboratory techniques used for the detection of monoclonal protein.(4, 5, 7) The diagnosis often rests on the demonstration of monoclonal plasma cell infiltrates in the marrow as well as presence of lytic bone lesions, which are common in these patients.(1, 3) A cytoplasmic M-protein can be identified by immunohistochemistry in nearly 85% of these patients.(8) With increasing sensitivity of detection of M-protein, the diagnosis of true NSM is decreasing.(9) This is especially true with the introduction of immunoglobulin free light chain assay that is capable of detecting small elevations in monoclonal free light chain, which would otherwise have escaped detection by immunofixation techniques. Patients with NSM represent a unique problem in terms of following the response to therapy due to the lack of a detectable monoclonal protein in the serum or urine. (3, 4, 6) No significant differences have been observed in terms of response to treatment in this group of patients compared to secretory myeloma, while others have reported better survival in these patients(5).(3, 6, 10). Due to the lack of a detectable M-protein, identification of disease progression tends to be delayed in this group compared to those with secretory disease. Autologous stem cell transplantation (ASCT) has been shown to provide a survival advantage compared to conventional chemotherapy.(11, 12) Given the differences in the clinical characteristics, it is not clear how the results of ASCT in this group of patients compare with those with secretory disease. We studied data reported to the CIBMTR, and compared the outcome of patients with NSM to a matched group of patients with SM.

PATIENTS AND METHODS

Patients were identified from among those undergoing an ASCT for multiple myeloma, between 1989 and 2003, reported to the CIBMTR. Patients who were planned to receive a tandem second ASCT, whether actually performed or not, were excluded from the current analysis. CIBMTR datasheets and pathology reports were reviewed to confirm the diagnosis in non-secretory patients. All non-secretory patients fulfilled the diagnostic criteria defined as not having a detectable paraprotein (intact immunoglobulin or light chain) by electrophoresis/IFE in the serum or urine. A set of patients with secretory disease matched to the non-secretory MM patients with respect to several clinical features were identified as described below.

A propensity score was used to match pairs for pre-transplant disease characteristics between the groups by fitting a logistic regression model. A numerical propensity score for each non-secretory MM transplant was calculated using the variables of age at transplant, Durie-Salmon stage at diagnosis, sensitivity to pre-transplant therapy, time from diagnosis to transplant and year of transplant. Propensity scores ranged from 0.02 – 0.127. The aim was to identify four matches for each non-secretory patient for the matched pair analysis. Non-secretory MM transplant recipients (cases) were matched in random order to secretory MM transplant recipients (controls) with similar propensity scores. Initially, non-secretory MM transplant (case) patients were selected randomly. Any secretory MM transplant (control) patient with a difference in the propensity score of less than 0.027 = (0.127−0.02)/4 was considered a potential matched control. The matched control with the smallest difference in propensity score among all potential matched controls was selected. These steps were repeated among the remaining cases until four controls were identified for each of the cases.

The primary outcomes studied were treatment related mortality, disease relapse/progression, progression-free survival and overall survival. Patient-, disease- and transplant-related variables for the two study groups were compared using the conditional logistic regression method to adjust the matching pairs. Probabilities of progression-free survival and overall survival were calculated using the Kaplan-Meier estimator; transplant-related mortality and relapse/progression were calculated using cumulative incidence estimates. The log-rank test was used for univariate comparisons. Multivariate Cox proportional hazards regression model stratified on matched pairs was used to examine the outcomes between non-secretory MM and secretory MM transplantation. The variables listed in Table 1 except those used in the modeling of the propensity score were used to build the final model. Any risk factors found to be significant were adjusted in the final Cox model stratified on matched pairs. All p-values are two-sided.

Table 1.

Variables used in the propensity score calculation for matching

	Non-secretory		Secretory
Variable	N eval	N (%)	N eval	N (%)	P-value
Age at transplant, years	110		438		0.91
<30 y		1 (1)		2 (<1)
30–39 y		3 (3)		8 (2)
40–49 y		28 (26)		114 (26)
50–59 y		50 (45)		204 (47)
≥60 y		28 (25)		110 (25)
Durie-Salmon stage at diagnosis	110		438		0.48
I		11 (10)		37 (8)
II		24 (22)		97 (22)
III		75 (68)		304 (70)
Sensitivity of myeloma prior to transplant	110		438		0.88
Sensitive		86 (78)		341 (78)
Time from diagnosis to transplant, m	110		438		0.99
<12 months		85 (77)		332 (76)
≥12 months		25 (23)		106 (24)
Year of transplant	110		438		0.34
1989–1992		7 (7)		19 (5)
1993–1996		18 (17)		80 (18)
1997–2000		36 (35)		189 (43)
2001–2003		42 (41)		150 (34)

RESULTS

The final dataset included 110 non-secretory MM HCT recipients and 438 secretory MM transplant (108 cases were found in 1–4 matches and 2 cases in 1–3 matches), selected from among 1771 patients reported to the registry during this time period, who had the variables available (total of 2219 patients reported during this time period). This yielded the 548 patients in the current analysis. At the time of the last follow up, 57 (52%) and 185 (42%) patients were alive in the non-secretory and secretory MM group respectively. The median follow-up of survivors was 61 months for the non-secretory MM transplant recipients and 68 months for the secretory MM transplant recipients.

Baseline characteristics

The propensity score based matching of cases with the controls resulted in the two groups of patients being well matched with respect to the six variables used for development of the scores as shown in Table 1. The two groups were comparable in terms of the baseline demographics as well as most of the disease characteristics from the time of diagnosis (Table 2). The non-secretory group had a higher proportion of patients with plasmacytomas identified at the time of diagnosis (38% vs. 10%) and was less anemic (median hemoglobin 13 gm/dL vs. 11 gm/dL) compared to the secretory myeloma group. The groups were otherwise comparable in terms of prognostic factors such as renal function, and serum LDH. The clinical characteristics were again fairly similar between the two groups at the time of SCT (Table 3). As at diagnosis, the secretory group had a lower serum albumin compared to those with non-secretory MM. More patients in the non-secretory group had received radiotherapy prior to the transplant, likely a reflection of the higher proportion of patients presenting with plasmacytomas. The groups were well matched in terms of the graft type (marrow vs. peripheral blood) numbers of nucleated cells collected and whether any graft purging was performed. The groups were also comparable in terms of the number and type of chemotherapy regimens they had received prior to the stem cell transplant.

Table 2.

Baseline characteristics at diagnosis of patients who underwent autologous transplantation for multiple myeloma by secretory status, reported to the CIBMTR between 1989 and 2003

	Non-secretory		Secretory
Variable	N eval	N (%)	N eval	N (%)	P-value
Demographics
Number of patients		110		438
Number of centers		58		85
Age, median (range), years	110	55 (28–74)	438	54 (23–74)	0.64
Male sex	110	60 (55)	438	264 (60)	0.25
At Diagnosis
Plasmacytoma at diagnosis	108	41 (38)	425	41 (10)	<0.001
Plasma cell (%) in bone marrow at diagnosisa	60	43 (0–100)	335	49 (0–100)	0.04
International Staging System at diagnosis	45		241		0.22
I		24 (54)		99 (41)
II		15 (33)		89 (37)
III		6 (13)		53 (22)
Immunochemical subtype of myeloma	NA		438		---
IgG				273 (63)
IgA				82 (19)
IgD				6 (1)
Light chain				76 (17)
Not specified				1 (<1)
Hemoglobin at diagnosis <10g/dL	95	14 (15)	400	154 (39)	<0.001
Albumin at diagnosis < 3.5 g/dL Creatinine at diagnosis	77	14 (18)	334	137 (41)	<0.001
>1 mg/dL	82	46 (56)	356	203 (57)	0.96
>1.5 mg/dL	82	16 (20)	356	96 (27)	0.11
>2 mg/dL	82	12 (15)	356	62 (17)	0.49
β2-microglobulin level at diagnosis ≥3.5 mg/L	52	16 (31)	256	90 (35)	0.62
LDH at diagnosis > upper limit of normal	51	9 (18)	224	39 (17)	0.65
Serum calcium at diagnosis >10.5 mg/dL	85	19 (22)	342	70 (20)	0.79

^aAmong patients who did not report extramedullary plasmacytoma at diagnosis.

Table 3.

Pre-transplant characteristics of patients who underwent autologous transplantation for multiple myeloma by immunochemical subtype, reported to the CIBMTR between 1989 and 2003.

	Non-secretory		Secretory
Variable	N eval	N (%)	N eval	N (%)	P-value
Karnofsky score pretransplant	108		419		0.33
<80		11 (10)		49 (12)
80		29 (27)		126 (30)
90–100		68 (63)		244 (58)
Time from diagnosis to first chemotherapy, median (range), months	106		421		0.77
<6 months		97 (92)		388 (92)
6–12 months		3 (3)		14 (3)
>12 months		19 (5)		19 (5)
Prior radiotherapy	110	49 (45)	437	134 (31)	0.005
Prior chemotherapy regimens	109		437		0.94
MP ± others		21 (19)		91 (21)
VAD ± others		68 (62)		261 (60)
Cyclophosphamide ± others		13 (12)		66 (15)
Corticosteroids ± others (no Cy, no melphalan)		7 (7)		10 (2)
Other		0		8 (2)
None		0		1 (<1)
Number of lines of chemotherapya	108		422		0.19
0		5 (5)		6 (1)
1		64 (59)		259 (61)
2		31 (29)		104 (25)
>2		8 (7)		53 (13)
Time from last chemotherapy to transplant, median (range), months	107		429		0.07
<6 months		94 (88)		343 (80)
6–12 months		11 (10)		70 (16)
>12 months		2 (2)		16 (4)
Disease status prior to transplant	105		414		0.50
Complete/ partial remission		83 (79)		325 (79)
MR/ NR/ stable disease		18 (17)		67 (16)
Relapse/Progressive		4 (4)		22 (5)
Lytic bone lesions anytime prior to transplant	102	94 (92)	352	318 (90)	0.30
LDH prior to transplant > upper limit of normal	85	32 (38)	347	115 (33)	0.70
Albumin prior to transplant < 3.5 g/dL	94	24 (26)	403	148 (37)	0.04
β2-microglobulin level prior to transplant ≥3.5 mg/L	67	8 (12)	254	51 (20)	0.10
Total number of nucleated cells (× 108/ kg) (PBSC±BM only)	80		351		0.79
< 2		15 (19)		58 (17)
2 – 5		22 (27)		89 (25)
> 5		43 (54)		204 (58)
Conditioning regimen	110		438		0.25
Melphalan only		55 (50)		236 (54)
Melphalan + TBI ± others		8 (7)		61 (14)
Melphalan + others (not TBI)		25 (23)		44 (10)
TBI ± others (not melphalan)		4 (4)		21 (5)
Othersb		18 (16)		76 (17)
Graft type	110		438		0.27
PBSC		102 (93)		406 (93)
Purging	110	5 (5)	436	29 (7)	0.41
Total number of transplantsc	110		438		0.32
1		106 (96)		411 (94)
2 or more		4 (4)		27 (6)
Planned post transplant treatment	109	47 (43)	433	227 (52)	0.09

Abbreviations: BU = busulfan; CY = cyclophosphamide; MOS = months; MR = minimal response; NR = no response; EVAL = evaluable.

^aExcludes stem cell priming

^bOther conditioning regimens were: Bu+Cy±others (18 cases in the non-secretory group and 67 in the secretory group) and other combinations (9 cases in the secretory group).

^cPatients who had a planned second transplant (whether they received their 2^nd transplant or not) were excluded from the analysis.

Outcome after stem cell transplant

Given the potential ambiguity in response assessments in patients with non-secretory myeloma, we focused our outcome comparisons on progression free survival and overall survival. The median progression free survival from SCT was 30 months (95% confidence interval (CI), 23–35) among the non-secretory patients compared to 23 months (95% CI, 19–27) for the secretory patients (P_logrank =0.05) (Figure 1). Similarly the median overall survival from SCT was 62 months (95% CI, 50–68) for the non-secretory patients compared to 50 months (95% CI, 42–57) for the secretory patients (P_logrank =0.26) (Figure 2). The median overall survival from diagnosis was similar for the two groups; 69 months (95% CI, 54–76) among the non-secretory patients compared to 59 months (95% CI, 53–69) for the secretory patients (P_logrank =0.34). In a multivariate analysis incorporating factors not used in the propensity score model, no differences were seen between the secretory and the non-secretory group (Table 4).

Figure 1.

Demonstrates the progression free survival following single autologous stem cell transplantation in patients with non-secretory or secretory myeloma. No statistically significant difference is seen between the two groups (log rank test).

Figure 2.

Demonstrates the overall survival following single autologous stem cell transplantation in patients with non-secretory or secretory myeloma. No statistically significant difference is seen between the two groups (log rank test).

Table 4.

Relative risks of relapse, treatment-related mortality, treatment failure and mortality with non-secretory vs. secretory multiple myeloma after ASCT in multivariate analysis.^*

Outcome event	Relative Risk (95% CI)	P-value
Treatment-related mortality
Non-secretory	1.00a
Secretory	1.20 (0.45 – 3.23)	0.72
Relapse/progression
Non-secretory	1.00a
Secretory	0.97 (0.68 – 1.39)	0.88
Treatment failure (death or relapse/progression)
Non-secretory	1.00a
Secretory	1.09 (0.79 – 1.52)	0.59
Mortality
Non-secretory	1.00a
Secretory	1.16 (0.80 – 1.68)	0.44

^aReference group

^*Multivariate analysis results are based on fitting a Cox model stratified on matched pairs adjusted for covariates not considered in the propensity score.

There was no difference in the treatment related mortality following SCT between the two groups (Figure 3, Table 4). The probability of treatment related death was 5–8% in both groups. At the time of the analysis, 48% and 58%, in the non-secretory and the secretory groups respectively, had died. Relapsed disease was the cause of death in the majority of patients (75%) in each group, with infections and organ damage being other common causes. There were two patients who had an allogeneic stem cell transplant following failed ASCT, and who had died due to graft versus host disease.

Figure 3.

Demonstrates the cumulative risk of treatment related mortality following single autologous stem cell transplantation in patients with non-secretory or secretory myeloma. No statistically significant difference is seen between the patient groups (log rank test).

DISCUSSION

Non-secretory myeloma, characterized by the absence of a monoclonal protein in the serum or in the urine, constitutes 1–5% of all myeloma and presents unique diagnostic and monitoring challenges.(1–7) Randomized trials have generally tended to exclude patients with NSM since they do not have easily measurable disease. In the few trials that did include patients with NSM (Child et al NEJM, 2003 =4%, Segeren et al, Blood 2003 =5% and Fermand et al, JCO 2005 =1%), the outcome is not clearly reported or compared with SMM.(12–14) In the majority of the patients with non-secretory myeloma, immunoflourescence studies reveal monoclonal protein in the cytoplasm of the plasma cells suggesting defects in secretion or rapid degradation of the protein as a common mechanism (“non-secretors”).(4, 5, 8) Mutations involving the immunoglobulin chains can also result in the inability of cells to secrete the protein.(15–17) Only a small proportion of patients after detailed plasma cell analysis fail to demonstrate any evidence of immunoglobulin heavy or light chain production.(18, 19) With the advent of more sensitive assays to detect circulating immunoglobulin free light chains, true NSM is becoming a rarity.(9, 20, 21) Many of these patients thought previously to have non-secretory disease will have detectable abnormalities in the light chain and may be or appropriately termed hyposecretory disease. In a study of 28 patients with “non-secretory” myeloma, 19 were found to have elevated free light chains with abnormal ratios(9). Several of the remaining patients had abnormal ratios with suppression of one of the light chains.

Several small patient series have described the presentation and clinical course of NSM patients. They tend to be younger by an average of 6 years compared to those with secretory myeloma, a feature that cannot be examined in the current study given the matching characteristics.(3) As in the current study, these patients tend to have better hemoglobin levels, with similar levels of renal function and prevalence of metabolic abnormalities such as hypercalcemia.(1, 3, 4, 6) Assessment of response to therapy remains a challenge and often depends on symptomatic improvement, reduction of plasmacytomas or bone marrow plasmacytosis, or correction of anemia and hypercalcemia. Published literature suggest comparable response rates in these patients compared with those with secretory disease.(4) Other studies have suggested a more favorable outcome with conventional therapy for non-secretory myeloma.(3, 6) Better outcome in some of these studies may reflect a detection bias as these patients may be identified earlier due to symptoms related to plasmacytomas but could also be due to differences in the disease biology. While recent studies have highlighted potential molecular mechanisms underlying non-secretory myeloma, data regarding the role of new agents is lacking. Information regarding the efficacy of ASCT in patients with NSM is very limited. Terpos et al reported improved progression free survival in a small sample of 6 NSM patients undergoing SCT compared to 121 patients with secretory MM(22).

Given the challenge of detecting disease relapse in NSM patients, PFS may not represent an ideal endpoint to compare outcomes. In the current study, we were not able to discern statistically significant improvement in PFS (p=0.05) among patients with non-secretory myeloma. The borderline p value likely represents the delay in detecting relapse in NSM patients (where relapse is defined by marrow plasmacytosis as opposed to biochemical criteria) rather than a true difference in the time to progression, since there was no difference in the overall survival after ASCT between the groups. While the reported mortality rate is higher than that seen in most single institution series or randomized trials, it was comparable between the groups. These higher rates likely reflect the effect of a large number of centers reporting to the registry as well as the long time period over which the current data was accrued. In the current study, by selecting a control group of patients undergoing ASCT in a contemporary period, balanced for most known prognostic factors, we provide conclusive evidence in favor of ASCT in patients with non-secretory myeloma. Clearly the data presented here is only applicable to single autologous SCT for this population since patients treated on planned tandem ASCT protocols were excluded from the analysis. This was confirmed by the small number of patients actually receiving a second stem cell transplant. It will be difficult to extrapolate the results from tandem ASCT strategies to this population, given the lack of benefit in the randomized trials for the subgroup achieving a VGPR, a response state not assessable in the NSM group.(23)

Data regarding measurement of serum free light chains, if available, would have provided valuable insight in the current study. Based on other studies, it is likely that a significant proportion of patients in the current cohort of NSM would have had detectable serum free light chains. Whether differences exist in the outcome between the “true” NSM and those with detectable free light chain abnormality by free light chain assays is certainly an important question and should be addressed in the future. This is relevant, since a previous study had suggested that patients with light chain myeloma had inferior survival following ASCT compared to intact immunoglobulin G (IgG) or A (IgA) myeloma.(24) This could also reflect a higher prevalence of the unfavorable translocations among patients with elevated free light chains.(25)

In conclusion, this study represents the largest series reported on the clinical outcome of patients with NSM. Since most clinical trials in the past tended to exclude NSM and given the rarity of the condition such an analysis is only feasible using registry data. In the absence of randomized trials including these patients, these data represents the best evidence supporting similar treatment approaches for this group of patients. With increasing use of the free light chain assay, the majority of patients with NSM are expected to have detectable light chain abnormalities thus making them oligosecretory rather than “truly non-secretory”. Clinical as well as biological differences between the groups of secretory, oligosecretory and truly non-secretory subtypes should be the focus of future studies.