Economic Burden of Disease Progression in Multiple Myeloma Patients Receiving Stem Cell Transplant

Multiple myeloma (MM) is a rare hematologic malignancy associated with complications such as anemia, cytopenias, bone disease, and renal dysfunction. In the US, approximately 32,270 new cases and 12,830 deaths occur annually, with a 5-year survival of ~54%. Guidelines (NCCN and IMWG) recommend therapy initiation when symptomatic or based on malignancy biomarkers. For transplant-eligible patients, initial regimens commonly include a proteasome inhibitor (e.g., bortezomib), an IMiD (e.g., lenalidomide), and dexamethasone, with an emerging role for monoclonal antibodies and quadruplet regimens. Despite multiple therapeutic options and subsequent line strategies for relapsed/refractory disease, response likelihood and duration generally decrease by line of therapy, and the disease remains ultimately fatal for many patients. Prior research has quantified substantial costs in later lines among progressing patients, but evidence specifically for patients receiving transplant has been limited. This study aims to compare and quantify HRU and healthcare costs between MM patients with versus without progression among those who received SCT within 1 year of diagnosis and at least one line of systemic therapy, using inverse probability of treatment weighting to adjust for baseline differences.

Retrospective studies have documented high costs among MM patients receiving second or later lines, focusing on those who progress and require additional therapy. Hagiwara et al. reported among transplant-ineligible MM patients (LOTs initiated 2006–2016) that total annual healthcare costs were higher for patients with versus without progression, with differences of $25,920, $30,632, $47,320, and $19,769 in L1–L4, respectively. Evidence specifically focusing on MM patients who received transplant has been lacking. The present study applies a similar approach to transplant recipients, with adjustments appropriate for transplant-eligible populations and LOT identification.

Design: Retrospective incident-user cohort study using IBM MarketScan Commercial Claims and Encounters (CCAE) and Medicare Supplemental (MDCR) databases from 01/01/2006–09/30/2018. The databases include fully de-identified enrollment and adjudicated medical and pharmacy claims data. Population: Adults with confirmed MM (≥1 inpatient claim with ICD-9-CM 203.0x or ICD-10-CM C90.0, or ≥2 non-diagnostic outpatient claims 30–365 days apart), with ≥6 months continuous enrollment pre-diagnosis, who initiated a qualifying MM therapy (lenalidomide, pomalidomide, thalidomide, bortezomib, carfilzomib, ixazomib, daratumumab, elotuzumab, panobinostat, vorinostat) during the study period and received SCT within 1 year of diagnosis. Exclusions: MM therapy or unclassified antineoplastics prior to diagnosis; evidence of more than one proteasome inhibitor simultaneously; missing supply amounts that could not be imputed; missing demographics; and gaps in enrollment around LOT initiation where specified. LOT identification: Adapted from a published algorithm for transplant-ineligible patients and refined for transplant-eligible patients via expert clinical review. L1 initiation was defined as first MM therapy receipt before the first SCT. Tandem SCT was assumed if a second SCT occurred within 1 year of the initial SCT. L2 initiation was the first MM therapy after SCTs. From L2 onward, the next LOT began at the earliest of: (1) first claim for a new MM therapy not used in the treatment identification period; (2) first claim for any MM therapy after a >90-day gap post-cycle; or (3) for those receiving more than one type of PI during the period, the first date with a claim for the second PI. Analyses focused on L1–L3 due to small numbers beyond L3. Progression definition and index assignment: For each LOT, progression was assessed during the first 12 months after LOT initiation. Progression was defined as: advancement to the next LOT; evidence of bone metastasis, hypercalcemia, soft tissue plasmacytoma, skeletal-related events/fractures, acute kidney disease/renal failure while off-therapy; or evidence of death (based on discharge status, diagnosis codes, and hospice within 1 month of disenrollment; excluding disenrollment at study end). For progressed patients, the index date was 30 days prior to documented progression (to capture prodromal and terminal care). For non-progressed patients, index dates were randomly assigned to mirror the time-from-LOT-initiation distribution of progressed patients. Baseline period was 12 months pre-index; patients required ≥12 months continuous enrollment pre-index. Follow-up began at index and continued to end of study (09/30/2018) or end of continuous enrollment, whichever first; follow-up was truncated at unclassified antineoplastic claims or at switch from commercial to Medicare. Outcomes: Annualized HRU (ED visits, physician office/outpatient hospital visits, other outpatient visits, outpatient prescriptions, inpatient admissions, inpatient days) and costs (MM therapy [drug + administration], SCT, other inpatient hospitalization, other outpatient services excluding SCT/MM therapy, and other outpatient pharmacy). Costs reflect total paid (insurer + patient) and were adjusted to 2018 USD using CPI for medical care. Cost categories were assessed hierarchically. Analysis: Comparisons between progressed vs non-progressed used stabilized inverse probability of treatment weights (sIPTW) to adjust for baseline differences (index year, age, sex, region, plan type, Medicare status, comorbidities including Charlson index, and pre-index HRU/costs). Logistic regression estimated the probability of progression to derive ATE weights; stabilized using marginal probabilities. Patients were additionally weighted by follow-up duration. P values were from t-tests with sIPTW; 95% CIs via bootstrap (2000 replications). Expected cumulative total plan costs by progression status were estimated using Kaplan–Meier sampling average (KMSA), censoring at disenrollment; subgroup analyses assessed LOTs initiated 2006–2012 vs 2013–2018.

Cohort: Of 58,946 adults with MM, 30,906 received MM therapy; 5,309 received SCT within 1 year of diagnosis; 3,599 met inclusion after exclusions. In L1, 3,599 included 934 with progression and 2,665 without progression; matched pairs numbered 904 (L1), 482 (L2), and 244 (L3). Baseline balance: Before weighting, progressed patients had more comorbidity and higher pre-index HRU and several cost categories. After sIPTW, baseline characteristics were well balanced in L1–L2 (standardized mean differences <0.1) and mostly balanced in L3. HRU differences (progressed vs non-progressed): - Hospitalizations per year were higher with progression across LOTs: L1 difference 0.32 (95% CI 0.20–0.44); L2 0.97 (0.73–1.22); L3 0.88 (0.50–1.26). Length of stay similarly higher. - Outpatient visits per year were higher with progression across L1–L3. Annual cost differences (progressed minus non-progressed; 2006–2018): - Total costs: L1 $18,359 (95% CI $2,289–$34,429); L2 $87,055 ($61,778–$112,333); L3 $71,917 ($36,452–$107,382). MM-related components comprised most of the incremental totals (approximately 91% in L1, 94% in L2, 91% in L3). - MM therapy costs were lower with progression in L1 and L3 and similar in L2, potentially reflecting unplanned discontinuations. - Costs for SCT, other hospitalizations, and other outpatient services were generally higher with progression. Temporal subgroup (2013–2018 initiations): - Total annual incremental costs: L1 $46,024 (95% CI $20,647–$71,400); L2 $100,329 ($67,909–$132,748); L3 $101,942 ($56,762–$147,122). HRU differences (hospitalizations and outpatient visits) were consistently higher with progression across LOTs. Cumulative costs (KMSA): - At 1 year post-index (2006–2018), incremental cumulative total costs were $11,863 (L1), $68,414 (L2), and $98,507 (L3). Over later years, L1 with progression showed lower cumulative costs than without progression, whereas L2–L3 with progression maintained higher cumulative costs. In the 2013–2018 subgroup at 1 year, increments were $35,279 (L1), $67,664 (L2), and $110,252 (L3).

Among MM patients who received SCT and systemic therapy, progression within 12 months of LOT initiation was associated with substantially greater HRU (more hospitalizations, longer stays, more outpatient visits) and higher annual and short-term cumulative healthcare costs compared with similar patients without progression. This pattern held across L1–L3 and persisted in more contemporary treatment eras (2013–2018). The notable exception was lower MM therapy costs among progressed patients in several LOTs, plausibly due to unplanned treatment discontinuation triggered by progression-related events or death. The findings align with prior evidence in transplant-ineligible populations, extending the burden-of-progression paradigm to transplant recipients. These results underscore the economic impact of early progression and aggressive disease biology in MM and suggest that delaying progression could reduce downstream medical resource use and costs beyond drug expenditures.

In transplant-eligible MM patients who underwent SCT and received systemic anti-myeloma therapy, disease progression within 12 months of treatment initiation was associated with higher HRU and healthcare costs across the first three lines of therapy and across time periods. Therapies that delay progression may provide meaningful reductions in subsequent disease management costs, an important consideration in value assessments of innovative MM treatments.

Use of administrative claims data entails potential misclassification due to coding errors and limited clinical granularity (e.g., lack of laboratory and pathology data). Some patient costs may be unobserved due to clinical trial participation or patient assistance programs. MarketScan populations may not fully represent all US patients (e.g., underrepresentation of small/medium employers and Medicare Advantage enrollees). Despite sIPTW adjustment, residual confounding by unmeasured factors may persist. The LOT identification algorithm, adapted from prior work, is not clinically validated and may misclassify lines. Progression was inferred using proxies (advancement to next LOT, specific clinical events, or death), with unknown sensitivity and specificity.