Cost of Illness of HER2-Positive and Recurrent HER2-Positive Breast Cancer – A Danish Register-Based Study from 2005 to 2016

Background: Information and knowledge about cost of illness and labour productivity in patients with HER2-positive early-stage and metastatic breast cancer treated with trastuzumab is limited. The aim of this study was to estimate the direct and indirect costs associated with treatment of HER2-positive breast cancer among patients with early-stage and metastatic breast cancer, treated with trastuzumab, in a 10-year period after diagnosis. Materials and Methods: This study included all Danish HER2-positive breast cancer patients ( ≥ 18 years) treated with trastuzumab between 2005 and 2016 identied in the Danish national registers. Among this population, patients experiencing metastatic breast cancer were identied. For the study populations, we estimated total healthcare costs and indirect costs for one year prior to the breast cancer diagnosis and up to 10 years after diagnosis compared with a group of matched controls free of breast cancer. Results: We identied 4,153 HER2-positive breast cancer patients, whereof 27% were identied with metastatic breast cancer. During the follow-up period of 10 years, we estimated excess healthcare costs of EUR 115,000 among the total study population compared to controls; EUR 211,000 among patients with recurrence; and EUR 89,000 among patients without recurrence. Healthcare costs were found to be highest in the rst year after diagnosis and also peaked in the year after recurrence. Labour productivity was signicantly lower among patients with recurrence 10 years after breast cancer diagnosis compared with controls. Conclusions: In this study, we estimated the direct and indirect cost associated with HER2-positive breast cancer to be signicantly increased during the 10 years after diagnosis, specically among patients experiencing recurrence of breast cancer. to breast cancer in the study period. To reduce the risk of introducing bias, we limited the matching criteria to only age and highest obtained education.


Introduction
Breast cancer is the most frequent type of cancer diagnosed among women in European countries, constituting 20-25% of all incident cancer diagnoses [1]. In Denmark, the annual number of cases is approximately 4,700 (age-standardised rate 86.1 per 100,000 women) [2,3], and worldwide, more than 2 million women were diagnosed with breast cancer in 2018 [4]. The survival rate for breast cancer has increased during the past decade, which is likely attributable to advances in mammography screening and improved treatments of early-stage breast cancer [5]. However, approximately 20% of breast cancer patients experience recurrent or de novo metastatic breast cancer, which is usually considered an incurable condition.
About 10-15% of early-stage breast cancer patients have HER2-positive breast cancer [6]. Before the introduction of treatments targeting HER2-positive breast cancer, it was an illness associated with higher rates of disease recurrence and higher mortality than other breast cancer subtypes. However, new targeted therapies have signi cantly changed the treatment paradigm of patients with early-stage HER2-positive breast cancer [7]. Since 2006, one-year treatment (17 cycles) with trastuzumab in combination with chemotherapy has been offered as standard treatment in Denmark to women with early-stage HER2-positive breast cancer before or after surgery.
Although the various subtypes of breast cancer affect many thousands of women worldwide each year, information and knowledge about the cost of illness and labour productivity of patients with HER2-positive early-stage and metastatic breast cancer treated with trastuzumab is limited. Denmark offers unique options for performing studies on real-world data via access to high-quality, exhaustive register data. Therefore, we used the Danish national registers to investigate the association between HER2-positive breast cancer (with and without recurrence), extended healthcare costs and labour productivity.

Methods
A retrospective population-based study was designed to analyse healthcare costs and labour productivity among Danish women diagnosed with HER2-positive breast cancer in the period 2005 to 2016 who were treated with trastuzumab.

Data sources
Since 1968, all Danish citizens have been assigned a unique personal identi cation number, recorded in the Danish Civil Registration System (CRS) [8]. For all individuals, the CRS registers date of birth, gender, vital status, region of residence and family relationships. The CRS enables an identity-secure linkage of information among the Danish national registers. Patient-speci c data were collected from the Danish National Patient Register (DNPR), which contains information on all somatic hospitalisations in Denmark since 1977. Moreover, all outpatient activities, emergency room contacts and psychiatric ward contacts, including diagnoses and performed procedures, have been included in the DNPR since 1995 [9]. Information on all tariffs and unit cost estimates for each somatic hospital contact (hospital admissions and outpatient visits) was retrieved from the DNPR, which was available from 2002 onwards. The date of diagnosis of breast cancer was retrieved from the Danish Cancer Register, which contains data on the incidence of cancer and on tumour characteristics among the Danish population, dating back to 1943 [10]. The HER2 status of study participants was obtained from the Danish National Pathology Register, which contains information on all pathology examinations conducted in Denmark since 1997 [11]. Data on primary healthcare services and unit costs were retrieved from the Danish National Health Service Register for Primary Care [12]. Prescription medicine costs were collected from the Register of Medicinal Product Statistics. Acquisition and unit cost estimates were based on the market price, including patient co-payment and public reimbursement. Information regarding home care services was retrieved from the Register of Municipal services. Information of employment was retrieved from the DREAM Database, which is owned by the Danish Ministry of Employment. The DREAM database includes information on weekly labour market public transfer payments, i.e. unemployment bene ts or disability payments, for all Danish citizens since 1991. Individuals receiving such a payment are included in the database for the corresponding year, while the remaining members of the workforce are not included.
Information regarding highest obtained education was obtained from the Population's Education Register [13] and was de ned by the highest obtained education at the time of breast cancer diagnosis.

Study population
The study population was de ned as women ≥18 years diagnosed with incident HER2-positive breast cancer in the period between 2005 and 2016 who were treated with trastuzumab. All women with incident breast cancer were identi ed in the Danish Cancer Register using the ICD-10 code C50. In the Danish National Pathology Register, information on HER2 status was retrieved using SNOMED code T04 (breast) in combination with either SNOMED code F29603 (HER2 receptor overexpression) or SNOMED code FE13B5 (HER2 gene ampli cation). In the DNPR, treatment with trastuzumab was identi ed using treatment code BOHJ13. More than six treatments with trastuzumab in the year following diagnosis resulted in inclusion in the study population.
Recurrence of HER2-positive breast cancer Study participants were assigned to one of two subpopulations: patients who experienced recurrence of HER2-positive breast cancer within the study period and patients who did not experience recurrence within the study period.
Patients with recurrence of HER2-positive breast cancer were de ned as either patients who were registered with metastatic breast cancer at diagnosis (de novo breast cancer) (TNM-M code AZCD41 in the Danish Cancer Register), patients who received more than 17 treatments of trastuzumab in the 13 months after diagnosis, or patients who received at least six trastuzumab treatments within the rst year after diagnosis, at some point afterward discontinued treatment for at least six months, and then initiated trastuzumab treatment again.

Control population
One control for each case was randomly selected from the general population via the CRS, matched on age and highest obtained education as of the year of breast cancer diagnosis. The DNPR was used to ensure that all controls were unexposed to breast cancer in the study period. To reduce the risk of introducing bias, we limited the matching criteria to only age and highest obtained education.

Unit costs and outcome variables
Outcome variables for this study included healthcare costs, prescription medicine costs and labour productivity. Total healthcare costs included costs regarding healthcare services (primary and secondary somatic care) and home care services.
The individual labour productivity value was de ned as a product of the part of the year the individual was working and the gross average yearly wage for women, adjusted by the number of effective weekly working hours. The labour productivity value estimations included only individuals considered to constitute the workforce, i.e. those between the ages of 18 and 65.
Due to legal restrictions regarding access to detailed data on prescription medicines, it was not possible to include this information in the healthcare costs. However, an analysis of the overall costs of prescription medicines produced an estimate of any particular differences in costs between cases and controls.
All costs were set to the 2016 price level, and tariffs in both the Danish National Health Service Register for Primary Care and the DNPR were in ated using the combined price and wage index for healthcare services, according to the Danish Regions [14]. Prescription medicine prices were not in ated, as they uctuate, making price indices di cult to interpret. In the present study, all costs are reported in euros and assume the following exchange rate: EUR 1=DKK 7.5.

Statistical analyses
The study followed participants from one year prior to their index date, i.e. date of incident breast cancer diagnosis (1 January 2005 at the earliest), and until 10 years after index date, death, emigration, or end of follow-up (31 December 2016), whichever occurred rst.
Average individual healthcare costs and labour productivity were calculated on a yearly basis according to index date for cases and controls, respectively. Due to the large number of observations in all categories, a one-sample t-test was applied to determine signi cance in differences between healthcare costs among cases and controls.
A secondary analysis estimated the healthcare costs of recurrence of breast cancer. Thus, date of recurrence was de ned as the date of breast cancer diagnosis if the patient was diagnosed with metastatic breast cancer (de novo breast cancer), as the date of the 18th treatment of trastuzumab if the patient was identi ed with more than 17 treatments of trastuzumab, or as the date of re-initiation of treatment with trastuzumab after a six-month break. Healthcare costs three years after recurrence were estimated.
Finally, further analysis included an estimate of the individual average costs of prescription medicines among cases and controls.
Data management and statistical analyses were carried out using SAS statistical software (9.4; SAS Institute, Inc., Cary, NC, USA) on Statistics Denmark's research computers via a remote server.

Results
Between 2005 and 2016, 7,156 incident HER2-positive breast cancer patients were identi ed in the Danish National Pathology Register. Among those, 4,153 (58%) were treated with trastuzumab within the rst year after diagnosis (see ow chart, Figure 1) and thus were included in the study population. Of these, 1,109 (27%) women were identi ed as patients with recurrence of cancer and 3,044 (73%) women were identi ed as patients without recurrence.
Basic characteristics of the study population are presented in Table 1. Age and highest obtained education at diagnosis did not vary in correlation to the population with recurrence of breast cancer and the population without recurrence. However, there was some variation in region of residence.
Average healthcare costs Table 2 presents the annual average healthcare costs per person among patients with HER2-positive breast cancer treated with trastuzumab compared to matched controls free of breast cancer among the total study population, the population with recurrence of breast cancer and the population without recurrence of breast cancer.
In the year before diagnosis (year -1), we found no statistically signi cant difference between recurrence cases' and controls' average individual healthcare costs. The differences in total costs between cases in the total study population and controls, and between cases without recurrence and controls, were statistically signi cant (p=0.0209); however, the costs were only slightly higher among cases compared to the costs among controls (1.2 times).
In the year of diagnosis, the average healthcare costs were 25.9 (p<0.001) times higher among cases in the total study population compared to controls, 27.3 (p<0.001) times higher among cases with recurrence compared to controls and 25.3 (p<0.001) times higher among cases without recurrence compared to controls, corresponding to differences in actual costs of EUR 58,217, EUR 67,099 and EUR 54,910, respectively.
In subsequent years, the average healthcare costs remained statistically signi cantly higher among cases alive compared to controls alive in all study populations. In the 10 th year after diagnosis, the average healthcare costs remained signi cantly higher among the total study population and among patients with recurrence. Among all cases, the costs were 2.3 (p=0.003) times higher compared to controls in year 10, whereas the costs were 10.0 (p=0.002) times higher among cases with recurrence and 1.2 (p=0.402) times higher among cases without recurrence compared to controls.
Labour productivity Figure 2 presents the average individual labour productivity among HER2-positive breast cancer patients compared to controls who were free of breast cancer from year -1 to year 10. In the year prior to index date, cases and controls in all study populations did not differ signi cantly regarding labour productivity: p=0.180 among the total study population, p=0.99 among cases with recurrence and p=0.103 among cases without recurrence. For all three study populations, labour productivity was slightly higher among cases compared to controls.
Among the total study population, labour productivity decreased signi cantly in the year of diagnosis and remained statistically signi cantly lower during the entire follow-up period. Among the population of cases with recurrence, labour productivity was signi cantly lower throughout the study period; in year 10, labour productivity was 2.8 (p <0.001) times lower among cases compared to controls, whereas cases without recurrence showed labour productivity comparable to controls from year 9 and onwards (p=0.114).

Healthcare costs of recurrence
Among patients experiencing recurrence, 201 patients (5.5% of cases) were identi ed with metastatic breast cancer at the time of diagnosis (i.e. de novo breast cancer), while the rest experienced a recurrence after a mean time of 1.5 years. Table 3 presents average individual healthcare costs three years after recurrence. In the rst year after recurrence, the average individual cost increased to EUR 38,383 among cases, compared to EUR 2,129 among their matched controls. Thus, a second spike in healthcare costs was related to recurrence among breast cancer patients.

Prescription medicine
The average individual costs of prescription medicines are presented in Table 4. In the year prior to index date, the costs of prescription medicines did not vary signi cantly between cases and controls. In year 1 and year 3, cases had slightly higher costs of prescription medicines compared with controls. In the remaining years, however, we found no differences in costs of prescription medicines between cases and controls. During the entire study period, the average yearly cost of prescription medicine was EUR 343 for cases and EUR 330 for controls.

Discussion
Using the Danish national registers, we carried out the rst large-scale assessment of the association between HER2-positive breast cancer and healthcare costs related to recurrence and non-recurrence. Our analyses showed higher healthcare costs in the 10 years after diagnosis among women with HER2-positive breast cancer who experience recurrence. During the 10-year follow-up period, patients in the total study population had excess healthcare costs of EUR 115,000 compared to controls, patients with recurrence of breast cancer had excess healthcare costs of EUR 211,000 compared to controls, and patients without recurrence had excess healthcare costs of EUR 89,000 compared to controls. The average individual healthcare costs were higher in the years after diagnosis among breast cancer patients experiencing recurrence than among those not experiencing recurrence. This can partly be explained by the fact that approximately 5% of the study population was diagnosed with de novo breast cancer at incidence. Due to differences in methodology, study design and included costs -and thereby in the chosen perspective of the analyses -it is somewhat di cult to compare the results of the present study with cost estimates identi ed in the literature. A recent study by Sussell et al. [15] investigated the cost of metastatic HER2-positive breast cancer in a US setting. The authors found cumulative healthcare costs among patients with metastatic HER2-positive breast cancer to be $412,903 (~ EUR 302,000) during the three years after diagnosis, primarily driven by outpatient visits and HER2-targeted therapy drug costs. Although the estimated healthcare costs were higher than those estimated in this study, the primary drivers were found to be similar, as costs of outpatient contacts including medication dispensed at hospitals constitute 82% of the total costs in this study. The differences in healthcare costs can partly be explained by the differences in cost of trastuzumab in Denmark and the US. During the study period, the cost of trastuzumab has been 2.5 times higher in the US compared with Denmark. Assuming a similar cost of trastuzumab in Denmark would result in increased outpatient costs corresponding to nearly EUR 200,000 in the three years after diagnosis.
The strengths of our study include its retrospective register-based design, which included all Danish women diagnosed with HER2-positive breast cancer treated with trastuzumab between 2005 and 2016 with complete information on healthcare costs up until 2016. By design, the risk of biases related to participation, outcome surveillance or follow-up in our analyses is therefore miniscule. Furthermore, we were able to adjust for age and highest obtained education by matching the cases and controls on these factors, which limited confounding. Healthcare costs were similar among cases and controls in the year prior to index date, which is essential in studies focused on cost of illness. Finally, the number of patients receiving trastuzumab in the study period is consistent with the sales gures of trastuzumab in Denmark, indicating a high validity of the procedure codes in the registers.
However, this study also has a number of limitations that require consideration. Firstly, the phrase "recurrence of breast cancer" is not well de ned. The Danish national registers do not include information on recurrence, so we de ned "recurrence" by the use of trastuzumab and information on metastatic cancer at diagnosis. Therefore, we cannot rule out an element of miscategorisation of patients in the recurrence and non-recurrence groups, which might have in uenced our results. Moreover, the incidence of HER2 breast cancer and the number of patients treated with trastuzumab varied and was particularly lower in the beginning of the study period. This could indicate an irregular use of the SNOMED code for HER2-positive and the procedure code for trastuzumab in the beginning of the study period, although the numbers stabilised in the last part of the study period. Thirdly, it was not possible to include the costs of prescription medicine in the total healthcare costs due to data access limitations. However, prescription medicine costs did not seem to vary much between cases and controls during the study period, which indicates that their absence may not have impacted the results signi cantly.

Conclusion
In conclusion, our study suggests that the healthcare costs for patients with HER2-positive breast cancer remains signi cantly higher than the healthcare costs for controls in the 10 years after diagnosis. Moreover, there is a need for future studies investigating how to identify patients with recurrence, since this information is not de ned in the Danish national registers.

Declarations Ethical considerations
The study was register-based and complied with the regulations and instructions set up by the Danish Data Protection Agency (J. nr. 2014-54-0664). We used only anonymised data and present data only in aggregate and anonymous form. We neither contacted any study participants nor required any active participation from them. Ethics committee approval and written informed consent are not required for registry-based research according to Danish law.

Availability of data
The data that support the ndings of this study are available from Statistics Denmark's Research Service. However, restrictions apply to the availability of these data, which were used under license/authorisation for the current study and so are not publicly available. Additional data analyses are available from the authors upon reasonable request and with permission of Statistics Denmark's Research Service.

Consent for publication
Since, we used only anonymised data and present data only in aggregate and anonymous form, consent for publication is not required.

Competing interests
This study was supported by Roche Denmark. Maria Spanggaard and Jens Olsen are employees at Incentive, which is a paid vendor of Roche Denmark. Kenneth Forsstrøm Jensen is an employee at Roche Denmark. Michael Anderson did not receive any funding from Roche to conduct this study.
Maria Spanggaard, Jens Olsen, Kenneth Forsstrøm Jensen and Michael Anderson do not have any non-nancial competing interests.

Funding
This work was supported by Roche Denmark.

Authors' contribution
Maria Spanggaard, Jens Olsen and Kenneth Forsstrøm Jensen contributed to the study design, the interpretation of the results and the drafting of the manuscript. Michael Anderson contributed to the interpretation of the results and revision of the manuscript and gave clinical expert input. All authors have approved the nal version of the manuscript to be published and agree to be accountable for all aspects of the work.
Year -1 Year *T-test for differences in average healthcare costs between cases and controls.
Note: Healthcare costs included costs of primary care, hospital admissions, outpatient contacts and home care. *T-test for differences in average healthcare costs between cases and controls.