Regional differences in treatment and outcome for myeloma patients in Sweden: A population based Swedish myeloma register study

Abstract Background We wanted to evaluate if health care for multiple myeloma (MM) patients is equal in different regions of Sweden. Aim To study differences in survival for MM depending on health care region and early use of modern treatment. Methods and results Data from the Swedish Myeloma Register from patients diagnosed between 2008 and 2017 was used. Cohorts were defined by the six healthcare regions (labeled A–F) in Sweden and modern initial treatment was defined as including certain drug combinations. To adjust for time to treatment bias, survival analyses were performed also for patients alive 6 months after diagnosis. In all treated MM patients (n = 5326), we observed a superior overall survival (OS) for region A compared to all other regions (p < .01 for all respectively). After adjusting for time to treatment there was also a superior survival in the region with highest use of modern initial treatment (region A) compared to the regions defined in the study as having intermediate and low use (p < .01 for both). In patients receiving autologous stem cell transplantation (ASCT) a superior survival was observed for region A compared to all regions besides region B. Similar results were seen when adjusting for a time to treatment bias. In patients not receiving ASCT, 75 years or older and adjusted for time to treatment bias, a difference was noted only between region A and E (log rank p = .04, HR 1.2, CI 1.00–1.44, p = .06). In multivariate analyses including age, international staging system stage and time period of diagnosis, differences in survival remained for patients receiving ASCT between region A versus C, D, E and F (p = .01, p < .01, p < .01, p = .03). Conclusion We observed a superior survival in region A for patients receiving ASCT. Explanations may be higher usage of modern initial treatment or regional residual confounding. For patients not receiving ASCT, 75 years or older, differences in survival could be adjusted for.

It is characterized by a steep increase in incidence with advancing age. In Sweden the crude annual incidence is 7/100 000 inhabitants and the median age at diagnosis has been estimated to be 70 years for men and 73 years for women. 2 The survival for MM patients has increased since the introduction of high dose melphalan with autologous stem cell transplantation (ASCT) as part of standard treatment for younger patients. [3][4][5] The introduction of proteasome inhibitors (PI: s), [6][7][8][9] immunomodulators (IMIDS: s) [10][11][12][13][14][15] and the combinations with PI: s and IMIDS: s 16,17 have led to gradual improvement of outcome over time, as has also the use of the anti-CD38 monoclonal antibodies. 18 Maintenance treatment with PI: s and IMID: s has also further contributed to better outcome. 19 While improved survival has been demonstrated mainly in randomized clinical trials not including most myeloma patients, a dramatic improvement has been reported also in population based studies. [20][21][22] Data from both the Swedish Lymphoma and the Swedish Leukemia Registries have demonstrated differences in outcome depending on which healthcare region in Sweden patients belong to. 23,24 In myeloma, treatment inequalities in subgroups of the population in other countries have also been earlier described. 25,26 Although there are national consensus guidelines in Sweden for treatment of lymphoma, leukemia and myeloma worked out by representatives from each health care region, the implementation and interpretation of these guidelines may still vary.
As the goal is to provide an equal health care in the whole country, with all suitable treatment considered for all MM patients regardless of region and age, an evaluation of survival and treatment choice between the healthcare regions was considered necessary.
The aim of this study was to investigate possible differences in overall survival (OS) between healthcare regions in Sweden in relation to treatment with considerations to known confounding factors.

| METHODS
Patients diagnosed with symptomatic MM from 2008-01-01 to 2017-12-31 registered in the population based Swedish Myeloma Register 27,28 were included in the study, provided that they also had a 1 year follow up report. The definition of symptomatic MM in the register used international myeloma working group (IMWG) criteria at the time. [29][30][31] The study's explanatory variable of main interest was the health care region which the patient belonged to. We labeled these regions A, B, C,  Modern initial treatment was defined as bortezomib in combination with either melphalan, cyclophosphamide, thalidomide, or treatment with lenalidomide, pomalidomide, carfilzomib or daratumumab regardless of combination. In the data no discrimination could be done between no modern initial treatment and data missing. Consolidation/ maintenance treatment was included as a separate variable. To adjust for differences in delay of reporting between the regions, only treatments started within a year from MM diagnosis were included. Later relapse treatment was not evaluated. Patients who did not receive treatment were excluded. Treatment is interpreted as intention to treat analyses in the study. R software was used for statistical analyses. 32,33 The study was approved by the ethical committee of Stockholm County (Dnr 2018/60, with approved amendment to Etikprövningsnämnden Dnr 2020-00394).

| RESULTS
Overall, in all six regions 5326 patients with symptomatic myeloma were analyzed after exclusion of 250 patients who did not receive treatment. The coverage for the Register during 2008-2017 compared to the Swedish cancer registry was 99.2% for the report at diagnosis. Coverage for the one-year report compared to reported diagnosis of MM to the register was lower in region A (82.6%) compared to the rest of the regions (95.2%-99.2%). Age adjusted incidence for all symptomatic myeloma was with 95% CI: 6.3 (5.7-7.0), 6.1 (5.3-7.0), 6.0 (5.5-6.6), 5.9 (5.2-6.5), 6.4 (5.8-7.1) and 7.3 (6.4-8.2) per 100 000 for region A, B, C, D, E and F respectively. The proportion of smoldering myeloma was 18% overall, varying from 14% in region F to 23% in region B. The use of modern initial treatment, as defined above, differed between the regions (Tables 1 and S1). The highest percentage (66%) was seen in region A compared to other regions (45%-52%). The percentage of patients treated with ASCT also differed with highest percentage (37%) in region A compared to the rest (26%-33%).

T A B L E 1 Patient characteristics, all treated patients with multiple myeloma
Consolidation/maintenance treatment was most frequent in region F (15%) and least frequent in region A (8%). In region A, 91% were alive  (Tables S2-S4).

| Survival in all patients
In the whole group of MM patients who received treatment, a significant superior OS for region A was observed when compared separately to region B, C, D, E and F (p < .01 for all regions respectively), ( Figure 1A). This was true also when including only patients alive Initial modern treatment appeared to increase in the regions over time ( Figure 3).  (Tables S5 and S6).

| Survival in ASCT patients
Multivariate analyses for patients alive after 6 months showed that a significant difference in survival remained for region A compared to C, D, E and F ( Table 2). Test of proportional hazards was non-significant (p = .14) for the global analysis but not for age (p = .01).
When patients alive after 6 months were divided by region and use of modern initial therapy there was a superior survival for the group with highest use compared to regions in the intermediate and low usage groups respectively (p < .01 for both), ( Figure 2B). This was not the case for the two subgroups not receiving ASCT (Figures S1 and S2).

| Survival in non-transplanted patients below 75 years of age
For these patients we did not see any significant difference in OS between region A and other regions. There was also no significant difference in OS when including only patients still alive 6 months after diagnosis ( Figure S3).  (Tables S5 and S6).
In multivariate analyses there were also no significant differences in survival between region A and the other regions ( Table 2).
Test of proportional hazards was non-significant for the global analysis (p = .07) but not for age (p = .04).

| Survival in non-transplanted patients 75 years of age and older
In this group of patients, we noted a superior OS for region A compared to region B, E and F (p = .02, p = .04, p = .02 respectively).  (Tables S5 and S6).
In multivariate analysis no significant difference in OS between region A and the other regions remained (Table 2) Test of proportional hazards was non-significant for the global analysis (p = .76).
Multivariate analyses including the variables modern initial treatment and consolidation/maintenance treatment were also performed.
These factors appeared significant for superior survival for the two subgroups of patients not treated with ASCT. In patients treated with ASCT however, the variables modern initial treatment and no consolidation/maintenance treatment did not turn out as significant (Tables S7-S9).

| DISCUSSION
In this nationwide study, using population-based data from 5326 patients with symptomatic myeloma in the Swedish Myeloma Register, we could demonstrate an improved survival over time and identify age at diagnosis, ISS stage but also use of novel drugs as initial treatment and consolidation/maintenance treatment as variables associated with improved survival. Importantly we also found regional differences. In the total group of MM patients, a superior OS was observed in region A when compared to all other regions separately.
The difference was evident in patients undergoing ASCT where region A had a significantly better OS compared to all regions except one, while in patients not receiving ASCT and 75 years or older, the difference in survival was no longer evident after adjusting for a possible time to treatment bias of 6 months.
The strength of this study is that it represents unselected and comprehensive data from the whole Swedish population including all age categories and stages for comparisons during a long period of time. Limitations include coverage for the one-year report which was less frequent in region A than in other regions, thus excluding more patients in that region from analyses in our study. This could possibly have biased the survival results and perhaps also the age distribution in that region. The fact that our analyses do not include data on comorbidity is also a limitation possibly implicating residual confounding. The results in univariate and multivariate analyses must also be interpreted with consideration for the limitations of a retrospective study.
Adapting modern initial treatment methods (new agents and the combination of these) was significant for superior survival in univariate analysis in all the subgroups. Use of modern initial treatment differed between the regions and was most common in region A. When  There were also differences in age between the regions. In the whole study group, percentage of patients with age 0-49 years, 50-59 years and 60-69 years were highest in region A. In patients receiving ASCT however, the age group 0-49 years, had highest percentage in region E, F and A. Although age was significant for survival in multivariate analysis in this group of patients, the differences in OS between region A and the other regions remained. Our conclusion is therefore that age alone was not sufficient to explain survival differences between the regions.
In conclusion we observed a better survival in region A for patients receiving ASCT compared to other regions. A higher usage of initial modern treatment or regional residual confounding may provide explanations. In patients not receiving ASCT, 75 years or older, survival differences were not evident after considering a time to treatment bias of 6 months.

ETHICS STATEMENT
The study was approved by the ethical committee of Stockholm