Reduction in late mortality among patients with Multiple Myeloma treated with Autologous Peripheral Blood Stem Cell Transplantation – a BMTSS Report

Abstract

Background:

Therapeutic practices for multiple myeloma (MM) have evolved, such that novel-agent-based therapy and autologous peripheral blood stem cell transplantation (aPBSCT) is the current standard. Whether cause-specific mortality has changed with time remains unclear.

Objective:

we examined late cause-specific mortality among patients with MM receiving aPBSCT from 1989-2014.

Study Design:

We conducted a prospective cohort study using participants enrolled in the enrolled in the Blood or Marrow Transplant Survivor Study. We created three eras to reflect changing MM therapy: <2000 (pre-thalidomide); 2000-2005 (thalidomide); 2006-2014 (lenalidomide). We used Kaplan-Meier techniques and Cox regression for examining all-cause mortality, and sub-distribution hazards models for cause-specific mortality.

Results:

1906 patients were followed for a median of 9.2y. Conditional on surviving 2y, the 10y-overall survival was 45%. The 10y cumulative incidence of myeloma- and non-myeloma-related mortality was 33% and 13% respectively. Multivariable analysis showed declining MM-specific mortality (sub-distribution hazard ratio [SHR]_2000-2005=0.80, 95%CI, 0.60-1.07;: SHR_2006-2014=0.46, 95%CI, 0.34-0.62; referent group: <2000), infection-related mortality (SHR_2000-2005=0.50, 95%CI, 0.29-0.85; SHR_2006-2014=0.35, 95%CI 0.21-0.60; referent group: <2000) and CVD-related mortality (SHR_2000-2005=0.45, 95%CI 0.20-0.99; SHR_2006-2014=0.41, 95%CI 0.18-0.93; referent group: <2000).

Conclusion:

While primary disease remains the major cause of late mortality, we observed a significant temporal decline in myeloma-, infection- and cardiac-related late mortality over past 25y.

INTRODUCTION

The past two decades have witnessed a sizeable improvement in survival among patients with Multiple Myeloma (MM).¹ This is particularly true for younger patients (≤65y) where 10y survival rates now exceed 35%.^2,3 With an increased lifespan, late complications of therapy and their contribution towards late mortality have become more relevant.

Novel-agent-based induction followed by autologous peripheral blood stem cell transplantation (aPBSCT) and post-transplantation maintenance are now the standard of care for MM patients.⁴ This approach has minimal early treatment-related mortality and leads to high response rates and superior survival.^5-7 However, subsequent neoplasms, notably therapy-related myelodysplasia/acute myeloid leukemia (t-MDS/AML) have been reported⁸, especially in the context of post-transplant maintenance with lenalidomide⁹. Further, aPBSCT survivors are at risk for infections and cardiovascular complications that may increase the risk of late mortality.^10,11

Two recent population-based studies show that primary disease remains the major cause of late mortality in MM patients.^12,13 However, these studies lacked details regarding therapeutic exposures, including receipt of aPBSCT and did not examine whether cause-specific mortality rates have changed with the changes in therapeutic practices over the past 25y. We address this knowledge gap by examining all-cause, and cause-specific mortality in a large cohort of 2y survivors of aPBSCT performed in MM patients between 1989 and 2014.

METHODS

Data Source and Study Participants

The Blood or Marrow Transplant Survivor Study (BMTSS) is a collaborative effort between City of Hope, University of Minnesota and University of Alabama at Birmingham (UAB), examining the long-term outcome of individuals who have survived two or more years after BMT, irrespective of disease status at entry into the cohort. To be included in the current report, patients had to have received aPBSCT for MM between 1989 and 2014, and survived at least 2y after transplantation, irrespective of their disease status. The 2-year time point was chosen to approximately coincide with a time point where the disease relapse risks are much lower and longer term complications from stem cell transplantation begin to emerge. This is particularly true for malignancies where stem cell transplantation may be curative such as leukemias, nevertheless, still represents a valid time point to study survivorship issues among patients with Multiple Myeloma.

Outcomes

We defined late mortality as death conditional on surviving the first 2y after aPBSCT. We captured information regarding vital status from medical records, National Death Index (NDI) Plus¹⁴, and Accurinct databases¹⁵. NDI Plus and/or medical records provided information regarding the date and cause of death through 12/31/2017. We extended the vital status information through 07/30/2019 using information from medical records and Accurinct databases. Two investigators (SG & SB) independently reviewed the available causes of death and reassigned them into primary and secondary causes using pre-established criteria (Supplement); we resolved any discrepancy by consensus.

Covariates

Information on demographic and clinical characteristics (age at aPBSCT, sex and race/ethnicity, year of aPBSCT and therapeutic agents used for preparative regimens), was obtained on all eligible cases from the institutional transplant databases. We arbitrarily created three transplant eras to reflect the changing paradigm in anti-myeloma therapy: a) 1989-1999 (pre-thalidomide era); b) 2000-2005 (thalidomide era); c) 2006-2014 (lenalidomide era). Information on pre-transplant therapeutic exposures was available for 53% of the cohort, and was included in a pre-planned sensitivity analysis.

Statistical Analysis

All survival analyses are reported conditional on surviving 2y following APBSCT. We computed 5y and 10y overall survival by using Kaplan-Meier techniques. We calculated the median follow-up time using reverse Kaplan-Meier method.¹⁶ We also computed all-cause late mortality for patients transplanted during the three transplant eras: 1989-1999, 2000-2005, and 2006-2014. We built Cox proportional hazard regression models to determine the association between transplant era and all-cause mortality, adjusting for age at aPBSCT, sex, race/ethnicity, and conditioning regimen. We computed 5y and 10y cumulative incidence of myeloma-related mortality as well as non-myeloma-related mortality, using other known causes of death as a competing risk. We used proportional sub distribution hazards model (Fine-Gray) for competing risks for identifying the association between cause-specific mortality and transplant era, adjusting for age at aPBSCT, sex, race/ethnicity, and conditioning regimen.

We used standardized mortality ratio (SMR) to quantify the risk of death in this cohort as compared to age-specific (5y intervals), sex-specific and calendar year-specific mortality rates in the US general population. The expected number of deaths was calculated by multiplying the number of person-years in each stratum by the corresponding US mortality rates, obtained from Centers for Disease Control and Prevention.¹⁷ Ninety-five percent confidence interval (CI) of the SMR was calculated using the Poisson regression method described by Vandenbroucke.¹⁸ We computed person-years at risk from 2y post-aPBSCT to date of death or date of censoring (July 30, 2019), whichever came first. We calculated SMRs for all-cause as well as cause-specific mortality.

For patients with available data, we categorized patients with/without receipt of novel agents (proteasome inhibitor and/or an immunomodulatory agent) during pre-transplant period. We conducted a sensitivity analysis to study the impact of pre-transplant therapeutic exposures on all-cause mortality including patients with available pre-aPBSCT therapeutic exposures.

Two sided tests with an alpha level of <0.05 were considered statistically significant. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc.) and STATA 13.0 (StataCorp LLC).

RESULTS

Overall, 1906 patients met eligibility for this analysis (eFigure 1). We summarize the demographic and clinical characteristics of the patient population in Table 1. Median age at aPBSCT was 58y (Interquartile range, [IQR], 52-63y); 58% were males and 58% were non-Hispanic whites. Majority of patients (71%) received aPBSCT after 2006, and 77% received Melphalan as a single agent for conditioning. Across the three eras, we observed noteworthy differences in demographic and clinical characteristics. These included increasing age at transplantation, increasing proportion of African Americans, and increasing use of Melphalan as a single agent for conditioning in recent years. Among the 1017 (53%) patients with known pre-transplant therapy, we observed a decline in the use of cytotoxic chemotherapy regimens and increasing use of novel agent-based therapies (Table 1).

Table 1:

Characteristics of Study Participants – Overall and by Study Era

Variable	Overall	1989-1999	2000-2006	2006-2014	P value
N (%)	1906 (100%)	142 (7.5%)	417 (21.9%)	1347 (70.7%)
Age at first transplant in years
Median (IQR)	58 (52-63)	52 (47-56)	57 (51-63)	58 (53-64)	<0.001
≥60y	788 (41.3%)	27 (19%)	156 (37.4%)	605 (44.9%)	<0.001
Sex (n, %)
Male	1095 (57.5%)	91 (64.1%)	240 (57.6%)	764 (56.7%)	0.24
Race/Ethnicity (n, %)
Non-Hispanic White	1106 (58.0%)	95 (66.9%)	228 (54.7%)	783 (58.1%)
African American	311 (16.3%)	16 (11.3%)	50 (12.0%)	245 (18.2%)
Hispanic	235 (12.3%)	20 (14.1%)	48 (11.5%)	167 (12.4%)	<0.001
Other*	107 (5.6%)	8 (5.6%)	23 (5.5%)	76 (5.6%)
Missing	147 (7.7%)	3 (2.1%)	68 (16.3%)	76 (5.6%)
Conditioning Regimen (n, %)
Melphalan	1461 (76.7%)	39 (27.5%)	312 (74.8%)	1110 (82.4%)
Others#	234 (12.3%)	100 (70.4%)	97 (23.3%)	37 (2.7%)	<0.001
Missing	211 (11.1%)	3 (2.1%)	8 (1.9%)	200 (14.8%)
Pre-transplant regimen (n, %)
Novel Agent$	786 (41.2%)	0 (0%)	93 (22.3%)	693 (51.4%)
Non-Novel agent	231 (12.1%)	68 (47.9%)	128 (30.7%)	35 (2.6%)	<0.001
Missing	889 (46.6%)	74 (52.1%)	196 (47.0%)	619 (46.0%)

N, number of patients; IQR, Inter Quartile Range

^*Other race includes American Indian/Alaska Native (n=4) Asian (n=89) and Unspecified (n=14)

^#Other conditioning regimens include Melphalan/Cytoxan/Busulfan (n=104), Cyclophosphamide/Total Body Irradiation (n=47), Melphalan/Total body Irradiation (n=42) and other conditioning chemotherapies (n=41)

^$Novel agent based therapy defined as inclusion of at least one immunomodulatory agent or proteasome inhibitor

Trends in all-cause mortality

After a median follow up of 9.2y (range 2.0-26.5y) from aPBSCT, we observed 893 (46.9%) deaths. The median overall survival of the entire cohort, conditional on surviving 2y after aPBSCT was 8.2y with a 5y- and 10y-survival rate of 63.4% and 44.9%, respectively. The 10y overall survival rates improved significantly across the three treatment eras (36%, 41% and 57% in <2000, 2000-2005 and 2005-2014 respectively, Figure 1A-B). In a multivariable Cox proportional hazard regression model, patients transplanted in the more recent eras had lower hazards of mortality (2000-2005: HR=0.81, 95%CI 0.63-1.04, p-value=0.10; 2006-2014: HR=0.63, 95%CI 0.48-0.82, p-value=0.001; referent group <2000; p_trend <0.001). Additional factors independently associated with increased risk of all-cause mortality included older age at transplant (≥60y: HR=1.21, 95% CI 1.05-1.4, p-value=0.007) and male sex (HR=1.19, 95% CI 1.04-1.4, p-value=0.01) (Table 2). In a sensitivity analysis including pre-transplant therapy information in the Cox regression model, we found that patients receiving novel-agent based pre-transplant therapy had superior overall survival (HR=0.77; 95%CI 0.6-0.98; p-value=0.03) (eTable 1).

Figure 1:

Survival Estimates for the 1906 patients with Multiple Myeloma who underwent autologous peripheral blood stem cell transplantation and survived first 2 years after transplant: Overall survival for the entire cohort (Panel A); Overall Survival stratified by transplant era (panel B); Cumulative incidence of Myeloma-related vs Non-Myeloma related mortality (Panel C).

Table 2:

Predictors of Overall Survival in the Study Population

Variable	Hazard Ratio (95% CI)	P value
Year of Transplantation *
1989-1999	REF
2000-2005	0.81 (0.63-1.04)	0.099
2006-2014	0.63 (0.48-0.82)	0.001
Age at Transplantation
<60 years	REF
≥60 years	1.21 (1.05-1.39)	0.007
Sex
Females	REF
Male	1.19 (1.04-1.36)	0.014
Race/Ethnicity
Non-Hispanic White	REF
African American	0.99 (0.81-1.21)	0.893
Hispanic	1.01 (0.81-1.27)	0.917
Other	1.49 (1.12-1.99)	0.006
Unknown	3.27 (2.65-4.03)	<0.001
Conditioning Regimen
Melphalan	REF
Others	1.02 (0.82-1.27)	0.862
Unknown	0.34 (0.23-0.51)	<0.001

^*Global p_trend < 0.001

CI, confidence interval.

Overall, this cohort was at a 5.27-fold (95%CI 4.9-5.65) higher risk of late mortality as compared to the US general population. The SMR was lower for patients transplanted in the more recent transplant eras (1989-1999: SMR=7.26; 95% CI 6.0-8.7; 2005-2005: SMR=5.74, 95%CI, 5.1-6.4; 2006-2014: SMR=4.58, 95%CI, 4.1-5.1). The SMR declined with follow-up time and after 10y from aPBSCT, the SMR was comparable to the general population (SMR=1.0; 95% CI 0.85-1.16) (Table 3).

Table 3:

Standardized mortality rates of study population as compared to reference US population

Variable	Person-Years	Events	SMR	95% CI of SMR	P value
Overall	11,616.74	761	5.27	4.9-5.65	-
Year of Transplantation
1989-1999	1365.51	113	7.26	6-8.68	-
2000-2005	3462.55	285	5.74	5.1-6.43	0.03
2006-2014	6788.68	363	4.58	4.13-5.07	<0.01
Age at first transplantation
<60y	7284.17	454	7.91	7.21-8.66	-
≥60y	4332.57	307	3.53	3.15-3.93	<0.01
Sex (n, %)
Male	6665.77	457	4.61	4.2-5.04	-
Female	4950.97	304	6.71	5.98-7.49	<0.01
Race/Ethnicity
NHW	7093.82	418	4.40	3.99-4.83	-
African American	1824.95	94	4.84	3.92-5.88	0.41
Hispanic	1411.86	80	5.70	4.54-7.04	0.03
Others*
Survival Time
0-5y	2763.94	514	15.31	14.69-15.96	-
6-10y	4873.28	205	3.50	3.28-3.74	<0.01
11-15y	2642.90	36	1.00	0.85-1.16	<0.01
16-20y	-	-	-	-	-

NHW, non-Hispanic white, CI, confidence interval, SMR, standardized mortality rate

^*SMR not reported for categories with less than 1000 person years at risk

Trends in cause-specific mortality

Myeloma-related Mortality:

Of the 786 patients with known causes of death (88% of all deaths), myeloma-related deaths were most prevalent (n=561, 71.4%). The cumulative incidence of myeloma-related mortality at 5y and 10y was 16.8% and 33%, respectively (Figure 1C). The 10y cumulative incidence of myeloma-related mortality declined over the transplant eras (1989-1999: 51%; 2000-2005: 46%, and 2006-2014: 24%, p-value <0.01) (Figure 2A). Multivariable analysis confirmed the temporal decline in the hazard of myeloma-related mortality (2000-2005: sub-distribution hazard ratio, SHR=0.80, 95%CI 0.6-1.1, p-value 0.1; 2006-2014: SHR=0.46, 95%CI 0.3-0.6 p-value <0.001; referent group <2000). Older age at transplantation was associated with increased risk of myeloma-related mortality (SHR=1.20; 95%CI 1.0-1.4, p-value 0.04); a trend towards worse mortality was also seen with use of conditioning regimen other than single agent melphalan (SHR=1.25; 95%CI, 0.96-1.62, p-value 0.097) (Table 4).

Figure 2:

Temporal Trends in cumulative incidence of late mortality among 1906 patients with Multiple Myeloma who underwent autologous peripheral blood stem cell transplantation and survived first 2 years after transplant: Myeloma-related mortality by transplant era (Panel A); Non-Myeloma-Related Mortality by transplant era (Panel B); Infection-related mortality by transplant era (Panel C); Cardiac-related mortality by transplant era (Panel D); and subsequent neoplasm-related mortality by transplant era (Panel E).

Table 4:

Risk Factors associated with Cause-Specific Mortality in the Study Population

Variable	Sub Hazard Ratio and 95% Confidence Interval
	Myeloma	Infection	Cardiovascular	Subsequent Neoplasm
Year of Transplant (referent group: 1989-1999)
2000-2005	0.80 (0.60-1.07)	0.50 (0.29-0.85)	0.45 (0.20-0.99)	0.66 (0.27-1.59)
2006-2014	0.46 (0.34-0.62)	0.35 (0.21-0.60)	0.41 (0.18-0.93)	0.54 (0.22-1.31)
Age at first transplant (referent group: <60y)
≥60y	1.20 (1.01-1.43)	1.58 (1.10-2.27)	2.70 (1.69-4.31)	1.33 (0.82-2.17)
Sex (referent group: females)
Male	1.02 (0.86-1.20)	1.16 (0.83-1.64)	1.55 (0.96-2.50)	1.67 (0.99-2.84)
Race/Ethnicity (referent group: non-Hispanic whites)
African American	0.88 (0.68-1.15)	1.41 (0.87-2.27)	2.04 (1.12-3.71)	1.30 (0.68-2.51)
Hispanic	0.94 (0.71-1.25)	1.44 (0.85-2.46)	2.20 (1.19-4.04)	1.23 (0.58-2.61)
Other	1.62 (1.15-2.27)	1.08 (0.46-2.57)	1.41 (0.50-3.93)	0.61(0.15-2.56)
Unknown	2.58 (1.99-3.35)	2.48 (1.49-4.15)	1.95 (0.93-4.07)	0.94 (0.37-2.42)
Conditioning Regimen (referent group: Melphalan)
Others	1.25 (0.96-1.62)	1.28 (0.78-2.10)	0.93 (0.40-2.14)	0.82 (0.35-1.93)
Unknown	0.36 (0.21-0.59)	0.15 (0.04-0.61)	0.28 (0.07-1.14)	0.17 (0.02-1.24)

Non-myeloma-related mortality:

The cumulative incidence of non-myeloma-related mortality was 6.2% (95%CI 5.2-7.3) at 5y and 12.7% (95%CI 11.0-14.5) at 10y. The 10y cumulative incidence of all-cause non-myeloma-related mortality did not change significantly across the transplant eras (1989-1999: 12.3%, 2000-2005: 12.7%, 2006-2014: 11.8%; p-value 0.40) (Figure 2B). The three most prevalent non-myeloma related causes of deaths included infection, subsequent neoplasms (SN) and cardiac disease, accounting for 18%, 10.4% and 8.8% of deaths respectively.

The 10y cumulative incidence of infection-related mortality was 8.2% (95%CI 6.9-9.6), and the SMR for infection-related mortality was 18.9 (95% CI 15.9-22.4). We noted a significant temporal decline in the 10y cumulative incidence for infection-related mortality (1989-1999: 14.9%; 2000-2005: 10.4%; and 2006-2014: 6.1%, p-value <0.01) (Figure 2C); corresponding SMR rates across the three eras were 35.10 (95% CI 23.28-50.37), 19.29 (95% CI 14.13-25.56) and 15.59 (11.94-19.91) respectively. This temporal trend was retained in multivariable analysis even after adjusting for age, sex, race/ethnicity and conditioning regimen (2000-2005: SHR=0.50, 95%CI 0.3-0.9, p-value 0.01; 2006-2014: SHR=0.35, 95%CI, 0.2-0.6, p-value <0.001; referent group: 1989-1999). Older age at transplant was also associated with higher hazard of infection-related mortality (SHR=1.58; 95%CI 1.1-2.3; p value 0.01) (Table 4). Further, in a sensitivity analysis, patients receiving novel-agent based pre-transplant therapy were at a lower risk for infection-related mortality (SHR=0.50; 95%CI, 0.3-0.9; p-value 0.02) (eTable 2).

The 10y cumulative incidence of cardiac-related mortality was 5% (95% CI 3.9-6.3%) and the SMR was 2.07 (95% CI 1.63-2.58). There was a modest decline in the 10y cumulative incidence of cardiac-related mortality over the three transplant eras (1989-1999: 5.7%; 2000-2005: 4.6%; and 2006-2014: 4.5%, p-value 0.09) (Figure 2D); corresponding SMR rates were 3.39 (95% CI 1.91-5.48), 1.87 (95% CI 1.2-2.74) and 1.90 (95% CI 1.34-2.61) respectively. This decline in cardiac mortality was statistically significant across the transplant eras in the multivariable analysis (2000-2005: SHR=0.45, 95% CI 0.20-0.99, p-value=0.03; 2006-2014: SHR=0.41, 95% CI 0.2-0.9, p-value=0.046; referent group 1989-1999) (Table 4). Additional variables associated with cardiac mortality included age ≥60y at transplant (SHR=2.70; 95%CI 1.7-4.3, p-value=0.01), race/ethnicity (African American: SHR=2.04; 95%CI 1.1-3.7; p-value 0.02; Hispanic: SHR=2.20; 95% CI 1.2-4.0; p-value=0.01; referent group, non-Hispanic white). Novel agent based pre-transplant therapy was not associated with cardiac mortality (eTable 2).

The 10y cumulative incidence of SN-related mortality was 3.8% (95% CI 2.9-4.8%); this was comparable to the general population with an SMR of 1.19 (95% CI 0.9-1.53). We did not observe a significant trend in 10y SN-related mortality across treatment eras (1989-1999: 4.3%; 2000-2005: 4.1%; and 2006-2014: 3.5%, p-value 0.28 (Figure 2E); corresponding SMR were 2.01 (95% CI 1.01-3.52), 1.20 (0.74-1.82) and 1.02 (0.68-1.47) respectively. Novel agent based pre-transplant therapy was not significantly associated with SN-related mortality (eTable 2).

DISCUSSION

Findings from our study suggest that primary disease continues to be the major cause of late mortality among MM patients undergoing aPBSCT. Other non-myeloma-related causes of death included infections, subsequent neoplasms and cardiac complications. Additionally, we noted a significant temporal decline in all-cause, as well as myeloma-related, infection-related and cardiac-related mortality, although no significant change in SN-related mortality was evident. Since many anti-myeloma therapies increase the risk of cardiovascular, renal, and infectious toxicities and are associated with the development of SNs, understanding causes of non-myeloma-related mortality has a potential to inform survivorship models and guide appropriate interventions.^9,19-22

While previous studies have described the risk of late mortality among aPBSCT recipients^12,23, none have focused solely on MM patients, especially in the context of the changes in therapeutic options that are being offered to MM patients, with enhanced myeloma-related efficacy yet unique risks of long-term toxicities.^2,4,9 Two population-based studies examined late mortality experienced by MM patients.^13,24 Both studies found that MM accounted for the vast majority of deaths, followed by deaths due to cardiac causes and SNs. Both studies demonstrated a decline in all-cause and cause-specific mortality. However, lack of treatment information limited these studies; thus, the studies included both non-transplant and transplant populations and it was difficult to understand the underlying causes of the decline in mortality.^13,24 We attempted to overcome these limitations by including a well-characterized cohort comprising of MM patients treated with aPBSCT over a period of 25y. Furthermore, this study uses the services offered by the NDI and Accurinct database, in addition to medical records to obtain a complete and comprehensive record of the deaths experienced by this cohort. In addition, we were able to study the impact of transplant conditioning regimen as well as pre-transplant therapeutic exposure on all-cause and cause-specific mortality in a sensitivity analysis.

Similar to prior studies ^2,13 we found that primary disease was the major contributor of late mortality among MM patients treated with aPBSCT. However, we found that the most common cause of late non-myeloma-related mortality was infection, followed by cardiac causes and SNs. Further, we found declining rates of all-cause as well as myeloma-related mortality during the study period largely reflecting the therapeutic advancements seen during the study period. During the study period, we noted increasing use of novel-agent-based pre-transplant therapy (most evident after 2005) as well as Melphalan-based conditioning (most evident after 2000), both of which were associated with improved (or a trend towards improved) all-cause and myeloma-specific mortality in our analysis. This is consistent with published literature with proven survival benefit from use of novel-agent based induction therapy²⁵, as well as single agent Melphalan based conditioning regimen²⁶. However, we noted improving all-cause and myeloma-specific survival across the three treatment eras even after adjusting for conditioning regimen and pre-transplant therapy. This suggests that additional factors such as improving supportive care and advancements in post-transplant therapeutic options (maintenance therapy, treatment for relapsed or refractory disease) in the recent years may be contributing to the improved outcomes.^27-31

We also observed declining rates of infection-related and cardiac-related mortality over the transplant eras. The former could be due to improvements in supportive care witnessed during the recent years.¹ Declining use of conventional cytotoxic chemotherapy and high-dose dexamethasone, for newly diagnosed as well as relapsed disease, may also be contributory, since the risk of infection is much higher among patients receiving cytotoxic chemotherapies (vs novel-agent-based therapies)³² and high doses of dexamethasone³³. However, newer therapies such as daratumumab and other immune based therapies are associated with increased risks of infections³⁴, and, future trends in infection related mortality should be studied in this population. Despite previous reports supporting increased risk of SNs among patients treated with immunomodulatory agents such as lenalidomide^8,9,22, we did not find a significant impact of pre-transplant immunomodulatory use on SN-related mortality. Another factor responsible for the declining cardiac-related mortality could be better patient selection for aPBSCT as other promising non-transplant therapies have become available in recent years for those considered eligible for transplant.^34,35

We need to interpret these findings in the context of its limitations. We did not have access to data on use of immunomodulatory agents during the post-transplant period. Specifically, we lacked data on the use of agents such as carfilzomib and post-transplant lenalidomide that has been shown to increase the risk of cardiac toxicity¹⁹ and SNs^9,22 respectively. However, lenalidomide maintenance therapy after transplant was widely adapted in the US only after 2012 following publication of two large phase III trials^36,37, hence majority of patients in our cohort may not have received maintenance lenalidomide therapy. Furthermore, lenalidomide therapy has been shown primarily to increase the risk of hematologic malignancies mainly therapy-related MDS and AML (t-MDS/AML)²², with a short latency period (23-29 months) and a high fatality rate (median OS 6.5 months)³⁸. As a result, by restricting our cohort to 2y survivors following aPBSCT, many of whom received lenalidomide in the immediate post-transplant period, we may have missed many patients with t-MDS/AML. Conversely, our follow-up may not be long enough to capture certain incident neoplasms (usually solid tumors) that tend to have longer latency period, particularly in the most recent transplant era^9,22. We did not have information on the disease status, and treatment received in the post-transplant period. Most importantly, we did not have details regarding the supportive care practices over the past 25y. In order to mitigate this, we divided our study population into study era that approximated with the changing landscape in myeloma therapy²⁴. We did not have information on all patients regarding comorbid conditions as well as behavioral risk factors such as hyperlipidemia, type II diabetes or smoking, which modify the risk of cardiovascular mortality in a myeloma survivor. Additionally, we conducted preplanned sensitivity analysis in a subgroup of patients with available data on pre-transplant therapeutic exposures. An increasing proportion of African Americans receiving transplant in recent years could be partially skewed by the addition of UAB as a contributing site in the latter years. Lastly, we were unable to look at comorbidity profile and socio-behavioral characteristics in our study cohort such as physical activity, dietary habits, smoking and alcohol consumption that may influence an individual’s risk of dying from competing causes of death.

To summarize, our findings show that prevention of disease recurrence remains one of the greatest challenges for MM patients treated with aPBSCT. Further, deaths due to infections, cardiac causes and SNs highlight the long-term health issues faced by aPBSCT recipients. Additionally, we note a significant temporal decline in all-cause as well as myeloma- and infection-related mortality during the study period and a trend towards decline in cardiac mortality. Our study provides evidence of the need for interventions that will ultimately improve outcomes and quality of life among patients with MM.

Highlights:

• Primary Disease is the major cause of late mortality in adults with Myeloma receiving autologous peripheral blood stem cell transplantation

• There has been a significant temporal decline in myeloma-, infection- and cardiac-related late mortality over the past 25y.