Patterns of radioiodine use for differentiated thyroid carcinoma in Brazil: insights and a call for action from a 20-year database

ABSTRACT Objective: This study aimed to explore the patterns of radioactive iodine (RAI) use for differentiated thyroid cancer (DTC) in Brazil over the past 20 years. Materials and methods: A retrospective analysis of the DTC-related RAI prescriptions, from 2000 to 2018, retrieved from the Department of Informatics of the Unified Health System (Datasus) and National Supplementary Health Agency (ANS) database was performed. RAI activities prescriptions were re-classified as low (30-50 mCi), intermediate (100 mCi), or high activities (>100 mCi). Results: The number of DTC-related RAI prescriptions increased from 0.45 to 2.28/100,000 inhabitants from 2000 to 2015, declining onwards, closing 2018 at 1.87/100,000. In 2018, population-adjusted RAI prescriptions by state ranged from 0.07 to 4.74/100,000 inhabitants. Regarding RAI activities, in the 2000 to 2008 period, the proportion of high-activities among all RAI prescriptions increased from 51.2% to 74.1%. From 2009 onwards, there was a progressive reduction in high-activity prescriptions in the country, closing 2018 at 50.1%. In 2018, the practice of requesting high-activities varied from 16% to 82% between Brazilian states. Interestingly, variability of RAI use do not seem to be related to RAI referral center volume nor state socio-economic indicators. Conclusion: In recent years, there has been a trend towards the lower prescription of RAI, and a reduction of high-activity RAI prescriptions for DTC in Brazil. Also, significative inter-state and inter-institutional variability on RAI use was documented. These results suggest that actions to advance DTC healthcare quality surveillance should be prioritized.


INTRODUCTION
D ifferentiated thyroid carcinoma (DTC) comprising papillary and follicular carcinoma account for the majority of thyroid malignancies, is responsible for almost 3.1% of all new diagnosis of cancers in 2018 worldwide (1). The increasing incidence of thyroid cancer, previously observed mainly in high-income countries, is also a real phenomenon in low-and middle-income countries (2). In Brazil, the statistics provided by the National Cancer Institute (INCA) for 2018 estimate an annual incidence of 9,610 new thyroid cancer cases, with an estimated incidence of 1.49 and 7.57 per 100,000 habitants/year for men and women, respectively (3).
Management of DTC is under continuous review and has changed substantially in recent years. Given a significant amount of new information in the field, physicians rely on Guidelines to embase care decisions. Guidelines serve as an evidence-based repository of knowledge, qualifying decisions and minimizing undesirable clinical variability, thus contributing to increased healthcare value. The DTC Guidelines of the American Thyroid Association (ATA) are under continuous review, and the last version has been released in 2015 (4). Nonetheless, despite the efforts to increase the uniformity of care, considerable variability of DTC management regarding radioiodine (RAI) use is well documented (5,6).
RAI utilization for DTC treatment dates from the 1940s (7,8) and as an adjuvant treatment can be prescribed in different activities, ranging from as low as 30 mCi to over 250 mCi. The goals of RAI therapy should be based on individual patients risk, and can be for three purposes: RAI remnant ablation facilitating the early detection of recurrent disease; RAI adjuvant therapy minimizing the risk of residual disease and RAI therapy for persistent and/or metastatic disease. Initially, RAI was recommended to all patients with DTC. However, more recent evidence, that applies especially to low-and intermediate-risk DTC suggests that the effectiveness of remnant ablation with 30 mCi was similar to treatment with 100 mCi (9)(10)(11). Additionally, two recent meta-analyses bring to light the risk of permanent non-malignant adverse effects on salivary, lacrimal, and gonadal dysfunction (12), as well as an increased risk of second malignancy related to cumulative RAI dosing (13). Additionally, the use of lower activities of RAI, instead of high activities, contribute to lowering health care system costs and environmental impact from radiation. Aligned with this evidence and given the lack of superiority of high activities of RAI for most DTC patients, RAI indications for DTC are being narrowed towards the use of lower activities in the last decades.
Few studies addressed country-wide patterns of DTC RAI treatment (14)(15)(16). This study aims to explore the temporal patterns of RAI use for diagnosed cases of DTC over the past 20 years in Brazil and its states.

MATERIALS AND METHODS
The Brazilian Unified Health System (SUS), a publicprivate mix, has three parts: the public subsector, the private subsector and the private health insurance subsector. Most of the Brazilian health services are provided by the public subsector, where services are provided and financed by the government at the federal, state and municipal levels to nearly 210 million inhabitants (17). Additionally, nearly 25% of the Brazilian population count on private health insurance healthcare, regulated by the National Agency for Supplementary Health (ANS). Hence, a retrospective study was performed in January and February 2020, in which the primary sources of information were the Department of Informatics of the SUS (Datasus, http://datasus.saude.gov.br) (18) and the ANS databases (http://ans.gov.br) (19).
Data were collected from 2000 to 2018, following Datasus codes for thyroid cancer managementrelated procedures (Supplemental Table 1). In the Datasus Outpatient Information System, the number of oncologic thyroidectomies and RAI medical prescriptions for different activities (30, 50, 100, 150, 200 and 250 mCi) are registered according to the state of origin, region (South, Southeast, Northeast, North, and Midwest) and for the entire country.
Data on RAI prescriptions for DTC in the private health insurance subsector was gathered from the ANS website by the state of origin, region and for the entire country (Supplemental Table 1 To explore the potential influence of socioeconomic factors on the patterns of RAI use across states in Brazil, we also gathered the most recent socioeconomic indicators data available, as follows: Human Development Index (HDI) from Brazil United Nations Development Programme (UNDP) for the year of 2017, Gross Domestic Product (GDP) per capita from Regional Account System: Brazil 2017 (Sistema de Contas Regionais: Brasil 2017), Gini Index from National Household Sample Survey (Pesquisa Nacional por Amostra a Domicílios -PNAD) for the year of 2013 and ANS coverage by state from the ANS website. The HDI uses distinct indicators to display a measure of average achievement in key dimensions of human development: a long and healthy life, being knowledgeable and have a decent standard of living. GDP is a metric that breaks down a region economic output per person and is calculated by dividing the GDP of a region by its population, aiming to assess wealth and prosperity. The Gini index measures the extent to which the distribution of income among individuals or households within an economy deviates from a perfectly equal distribution. Finally, the ANS coverage by state represents an indicator of the extent of the private health insurance healthcare availability for the population of each state.
The project was approved by the ethics committee of Hospital de Clínicas de Porto Alegre (CAAE 29670919.9.0000.5327/GPPG 2019-0764), and all authors vouch for the accuracy of the data and analyses.    To explore the variability in DTC RAI treatment among the Brazilian states, we show a 2018 snapshot of the RAI adjusted prescriptions for 100.000 inhabitants and the proportion of high RAI activities ( Figure 3). This data demonstrates that the practice of ordering high RAI activities varied from 16% in Pernambuco to 82% in Goiás and that RAI prescriptions adjusted to 100,000 inhabitants, varied from 0.07 in Mato Grosso do Sul to 4.74 in Rio Grande do Norte.

RAI activities in Brazil
We then explored whether socioeconomic indicators or RAI referral center volume contribute to the variability of RAI practices identified. We found no correlation between the number of RAI prescriptions/100,000 inhabitants and %HA-RAI with four socioeconomic indicators (HDI, GDP, Gini index and ANS coverage) (  We then looked at the number of RAI prescriptions per institution (Supplemental Table 2). In 2018, 11 (21%) institutions prescribed more than 100 inpatient RAI activities (High RAI volume institutions), another 11 (21%) between 50 and 100 inpatient RAI activities (Intermediate RAI volume institutions), and 31 (58%) less than 50 inpatient RAI activities (Low RAI volume institutions). We looked at the proportion of inpatient high RAI activities prescribed according to institution volume: high volume institutions prescribed 1079/2020 (53%) high RAI activities, intermediate volume institutions 559/848 (66%) high RAI activities, and low volume institutions 310/573 (54%) high RAI activities. Additionally, there was no correlation between the number of inpatient RAI activities prescribed by institution and the proportion of high RAI activities (R = 0.006). Taking together these findings do not support that RAI practices for DTC therapy accompany institutional RAI prescriptions volume.

DISCUSSION
Here we demonstrated that the use of RAI and high RAI activities for DTC in Brazil seems to be decreasing. Nonetheless, in contrast with current international clinical guidelines, the proportion of high RAI activities is still surprisingly high. Additionally, we documented a significant variability of coverage and patterns of care regarding RAI use for DTC between different Brazilian institutions and states. We also believe that data on DTC treatment should be better organized to make it readily available and allow its users to monitor and evaluate treatments and outcomes. Better access will facilitate a reduction in treatment heterogeneity and improve access to health care across Brazil. In recent years there is a growing consensus that many DTC patients do not benefit from RAI treatment, and that a considerable portion of DTC patients can be treated with low RAI activities, reducing treatmentrelated damage. Thus, the observed reduction in DTC RAI prescriptions in recent years shows that clinical practice has this trend indeed. Another aspect to highlight is that DTC RAI prescriptions and high RAI activities seem to decline and follow recent changes in guideline recommendations despite the evolving healthcare coverage in Brazil. Nonetheless, the proportion of high RAI activities still represents more than 50% of DTC RAI prescriptions across the country. Given the lack of benefit of high RAI activities to most DTC cases and the well-known adverse effects due to unnecessary increased exposure to radiation, the practice of ordering high RAI activities for DTC is worrisome. The unjustified use of high RAI activities increase the risk of radiation-induced benign and malignant complications and pose an economic burden to the health system and the environment (21). We understand that RAI therapy for persistent and/ or metastatic disease may be eligible to RAI activities higher than 100 mCi. However, most DTC cases are restricted to the thyroid gland (~70%), around 30% presents cervical lymph node metastasis and only ~5% distant metastasis (4). Thus, one should note that the indication of high RAI activities may apply to a small subgroup of DTC patients. Notwithstanding, still for this subgroup of advanced DTC, the optimal therapeutic activity remains uncertain and controversial.
Another aspect to consider is the disparities in Brazilian health care coverage. Some Brazilian states still do not have access to RAI therapy, so adequate DTC adjuvant treatment is not provided for a substantial portion of the affected cases. This problem is not restricted to the public subsector because RAI prescriptions are not registered both either SUS or ANS healthcare beneficiaries in some states. In this matter, our study underscores not only overuse of treatment strategies in some states, but also underuse of RAI for DTC all over Brazil. Within this context, our data documents considerable DTC variability of health care among Brazilian states. Several factors may explain these findings. First, the low quality of evidence on DTC RAI treatment indicates clinical practice variability, a well-described phenomenon documented in previous studies from other countries (5,6,22). Second, economic and regional issues are of particular interest in Brazil, since the heterogeneous access to diagnostic and treatment delineate decision pathways with different tradeoffs according to the particular contexts. Interestingly, however, DTC RAI variability of health care documented do not seem to be explained by socioeconomic indicators nor RAI center volume, indicating that other variables must come into play.
We should also comment on the limitations of this study. The Datasus and ANS databases, the best known national data available, are comprised of aggregated ecological data. Thus, as these systems were not designed as cancer registries, they are prone to bias. Notwithstanding, given that both Datasus and ANS systems are based on billing information, they are audited by competent authorities, what contributes for a minimal curadorie of the data. This fact is also a call for action to work towards a national consolidated DTC database, that encompasses individual patient and tumor characteristics, as well as diagnostic and therapeutic resources used, along with clinical outcomes. Also, the proposed metrics on RAI use are new, a fact that limits benchmarking with other healthcare scenarios and countries. The studied estimates relay on diagnosed cancer cases, and therefore may be prone to some level of detection bias. Taken together, these factors may result in less accurate estimates over the studied period. Thus, it would be interesting if our findings could be complemented by multi-institutional health record review based-study.
In summary, the observed estimates of RAI overand underuse, the practice of ordering high activities and the significant variability of provided health care among states determine the predominant pattern of RAI use for DTC in Brazil. Also, population-adjusted DTC RAI prescriptions, DTC RAI prescriptions adjusted to new cases or cancer-related thyroidectomies, and percentage of high activity RAI prescriptions emerge as potential valuable metrics to compare, monitor, and evaluate DTC RAI patterns of treatment. Our findings are descriptive but may serve as a call for action, highlighting significant opportunities to improve DTC care in Brazil, and suggesting that advanced surveillance of DTC healthcare quality and improvement of the national cancer registry system should be prioritized.