Wisely Frugal: Ensuring sustainable funding for novel cancer therapeutics using cost-effectiveness thresholds in resource limited settings

Introduction

Sri Lanka has an age adjusted annual incidence of 129 cancer cases per 100,000 population with nearly 32,000 new patients being diagnosed each year, according to data from the national cancer registry¹. However, the actual incidence might be considerably higher due to under-reporting of cases².

Cancer care in Sri Lanka is predominantly provided through its public funded state health system which is free at the point of delivery^2,3. Funded by general taxation, the state health system functions as a network of primary,secondary and tertiary care hospitals under the administrative control of the Ministry of Health^2,3. Clinical Oncology services are provided by 26 cancer centres located throughout the island^2,3.

Each year, the Ministry of Health of the government of Sri Lanka allocates around 210 million US$ for the procurement of drugs for hospitals under its purview⁴. In the beginning of 2022, Sri Lanka faced a foreign exchange crisis arising from years of imprudent macroeconomic fiscal policies aggravated by the Covid-19 pandemic⁵.

The crisis was devastating and almost led to a calamitous breakdown of Health services and procurement of pharmaceuticals imported from abroad proved extremely challenging⁵. Since virtually all oncology drugs are imported, the economic crisis posed a major threat to cancer care.

We performed a cost-effectiveness analysis of novel cancer drugs with a view to compiling a list of essential drugs and approved indications to help prioritise procurement and help mitigate the impact on cancer care.

Methods

Results

The average exchange rate for the year 2021 was 200 Sri Lankan Rupees per United States Dollar (US$)⁷. The per capita GDP for the year 2021 was US$ 3815. The cost of treatment per standard treatment course was obtained for 83 oncology drugs⁸. Conventional agents which were excluded from the analysis are shown in Supplementary Table S1 along with the cost per standard treatment course.

After exclusion of these drugs, 41 drugs spanning across 83 indications in the palliative setting and 6 drugs across 10 indications in the adjuvant setting were included in the analysis. A full description of the analysis including the clinical trials on which outcome data was obtained from is included in the supplementary appendix.

There were two drugs for which de-escalated treatment was of relevance. When analysing abiraterone, we considered a low dose treatment regimen of 250 mg with food and the standard dose of 1000mg on an empty stomach for each indication. Similarly for adjuvant trastuzumab we considered 6 months of treatment as well as 12 months of treatment.

Tables 1 and 2 list the highly cost-effective and cost-effective drug indications respectively along with the total annual cost of procurement for these drugs. Supplementary Tables S2 and S3 list the drugs and indications which are potentially cost-effective and not cost-effective respectively. Figure 1 shows the plot for cumulative annual procurement cost against per capita GDP per life year gained, while Figure 2 shows the same plot with the thresholds limited up to four times the per capita GDP per life year gained.

• Download figure
• Open in new tab

Figure 1 Cumulative cost of treatments of novel drugs in relation to cost-effectiveness thresholds based on per capita GDP per life year gained.

• Download figure
• Open in new tab

Figure 2 Cumulative cost of treatments of novel drugs in relation to cost-effectiveness thresholds upto per capita GDP times four per life year gained.

The total cost of treatment is US$ 6 million if a threshold of per capita GDP per life year gained (GDP x 1) was set and US$ 16.3 million if it was per capita GDP x 3 per life year gained. If the threshold is increased to GDP x 4, the total cost would rise to US$ 47.3 million. If no threshold was set, Sri Lanka’s health system would need US$ 295 million to fund these novel drugs.

Discussion

The allocation of funds for healthcare in a public funded state health system is determined by political authorities and is often influenced by the general macroeconomic situation of the country⁹. Once the allocation is decided, it is imperative that cost-effectiveness based thresholds be considered to ensure maximum benefit from the drugs procured by each health system¹⁰.

In this work we present a rational basis for prioritising procurement of supply of oncology drugs in Sri Lanka which we believe would find resonance with health care systems of other low and middle income countries (LMICs). We also provide further evidence for the validity of the cut-off of GDP x 3 per life year gained as the threshold for determining cost-effectiveness ^11,12.

Based on this threshold the budget for procurement of novel cancer therapeutics would be approximately 16.8 million US$ which would be around 9% of the total drug budget of the state health system. Increasing the threshold to GDP x 4 would nearly triple the amount of funding required, further validating the robustness of the WHO threshold of GDP x 3. Without cost-thresholds, the cumulative annual cost of the currently procured novel drugs would be nearly US$ 300 million assuming every eligible patient would receive treatment. This is around 1.5 times the total annual budget of all drugs in the state health sector. In the absence of well defined cost-effectiveness thresholds, a health system such as ours will be open to frequent shortages of essential drugs when the budgeted allocation is expended or if unforeseen factors result in a foreign exchange crunch.

In a time of financial crisis, the lower threshold of per capita GDP per life year gained (GDP x 1) can be used to prioritise procurement. As shown by our data, a total allocation of approximately US$ 6 million would be sufficient to ensure supply of these highly cost-effective drugs. Furthermore, defining a cost-effectiveness threshold would also provide an incentive for pharmaceutical suppliers to reduce the price of drugs thereby leading to cost savings. These savings could be channelled to more cost-effective treatment modalities such as radiotherapy and surgery. Indeed, studies have shown that the dearth of quality radiotherapy resources in Sri Lanka has adversely impacted on outcomes of potentially curative cancers^13,14. Investing in screening and streamlining early detection pathways may also lead to significant improvements in survival ^15,16,.

Except for ibrutinib which is not registered in Sri Lanka, all treatments mentioned in the World Health Organisation (WHO) essential drugs list were found to be cost-effective in our setting as well¹⁷. However, there were several other treatments that were not included in the WHO list that were found to be cost-effective in our study, which are listed in Box 1. Two such treatments, adjuvant osimertinib in resected in adenocarcinoma of the lung and adjuvant olaparib in early germline BRCA mutation positive breast cancer, are very recent developments ^18,19. Nevertheless, this underscores the importance of performing local cost-effectiveness assessments to take into account cost variations in different health systems.

Another salient finding of our work is the cost-savings that can be achieved by using lower doses of abiraterone for metastatic prostate cancer and a shorter duration of adjuvant treatment with trastuzumab in early breast cancer, both of which have robust evidence in the form of non-inferiority randomised clinical trials^20,21.

For abiraterone a lower dose of 250mg with food was shown to have equal efficacy in terms of biochemical response in castration resistant prostate cancer²⁰. The clinical equipoise can be safely extrapolated to the hormone sensitive phase of the disease as well and this is borne out by the inclusion of the lower dose option in the NCCN guidelines for both settings²².

Even though the landmark PERSEPHONE trial of more than 4000 patients proved non-inferiority for 6 months of adjuvant trastuzumab with 12 months of treatment, the oncology community has been slow to adopt this partly due to concerns with its subgroup analysis showing superiority of 12 months of treatment in patients receiving concurrent trastuzumab with chemotherapy^23,. However clinicians in LMICs such as ours would be well advised to opt for six months of adjuvant trastuzumab due to its substantial cost savings and the uncertain benefit of one year of adjuvant treatment to one year, which if at all, is likely to be marginal. The newer anti HER-2 monoclonal antibody Pertuzumab was not even remotely cost-effective either in the adjuvant or metastatic setting. Trastuzumab emtansine has not been used in the state health sector and we were therefore unable to perform an analysis of cost-effectiveness. However, unless very substantial price reductions are made these agents are unlikely to be cost-effective in our setting.

With regard to multiple drugs for the same indication, we found that abiraterone was substantially more cost-effective than enzalutamide in metastatic prostatic cancer both in its hormone sensitive and castration resistant phases. Similarly gefitinib and erlotinib were superior to osimertinib in the first line treatment of metastatic adenocarcinoma of the lung, while pembrolizumab was more cost-effective than nivolumab in metastatic melanoma.

Despite gaining approval for multiple malignancies, the immunotherapeutic agents Pembrolizumab and Nivolumab failed to reach the cost-threshold in almost all indications with the exception of metastatic melanoma, where it was potentially cost-effective.

It is unlikely that these drugs would be affordable in health systems of LMICs such as ours in the near future. However, encouraging results from recent trials exploring the efficacy of lower doses of these agents. provides some hope, and more studies in this space are an imperative need ²⁴.

Analysis of agents used in the first line treatment of Chronic Myeloid Leukaemia posed many issues since it is takes the form of a chronic disease entity where treatment extends beyond 10 years, When considering imatinib in the first line treatment of CML, its cost is cheaper than the comparator interferon-alpha and low dose cytarabine. Since the annual treatment cost per patient for imatinib was only 180 US$ we excluded it from this analysis since its cost-effectiveness is evident. Due to short follow-up in the clinical trials of first line treatment of CML, with the novel tyrosine kinase agents nilotinib and dasatinib, it was not possible to determine the survival gain accurately ²⁵. However the overall survival gain over imatinib is likely to be very modest in the first line setting and these agents are substantially more expensive than imatinib²⁵. Over a 10-year period nilotinib and dasatinib would cost US$ 53000 and 44347 more than imatinib, respectively.

For our analysis we only considered the direct cost of the drug. The cost of drug administration, investigations, staff costs etc. were not included. Furthermore, since data on quality of life was not available, we were compelled to consider life years gained as the outcome variable rather than quality adjusted life years. The paucity of data also meant that more robust cost-effectiveness analysis methods such as a Markov model could not be performed. As such, our data is likely to overestimate the benefit of treatment. More studies on evaluating quality of life and cost of treatment in the state health sector are required as a matter of urgency.

The inability to determine quality adjusted life year gains for novel therapeutics should not deter health systems such as ours from using more simple metrics when setting cost-effectiveness thresholds. As shown by our results even rudimentary cost-effectiveness analysis could provide valuable insights when making decisions on funding. Some data is better than none in this regard.

Our intention was to highlight the importance of cost-effectiveness thresholds in determining novel cancer drug procurement and usage in health systems such as ours. We believe this has been achieved by our work, notwithstanding the limitations mentioned above.

List of Tables and Figures