1 Introduction
Lung cancer remains the most prevalent cancer and the leading cause of cancer-related death in China, with 733,000 newly diagnosed cases and 610,000 deaths in 2015 [
1]. Approximately 85% (range 80% and 90%) of cases are represented by non-small cell lung cancer (NSCLC) [
2‐
4], and majority of these cases are advanced disease at the time of first diagnosis [
5]. Historically, the treatment for advanced NSCLC (aNSCLC) is rarely curative, and usually has a poor effect on increasing survival [
3], with a 5-year survival rate of 16% [
6]. Despite significant advances in targeted therapies, antiangiogenics, and immunotherapy for the first-line treatment of aNSCLC [
7‐
9], subsequent lines of therapy for patients who experience progression on/after first-line chemotherapy are generally limited [
10].
In the last 10 years, the introduction of immune checkpoint inhibitors (ICIs) into the second-line setting for NSCLC, particularly inhibitors of the programmed cell death protein 1 (PD-1) axis, has transformed the therapeutic landscape in this recalcitrant disease [
11]. In China, nivolumab is the first PD-1 inhibitor licensed as a subsequent agent for aNSCLC. The pivotal studies (CheckMate 017 and CheckMate 057 Phase III trials, CA209-003 Phase I study) [
12,
13], confirmed that nivolumab dramatically prolonged overall survival (OS), and had a more favorable safety profile compared with standard second-line treatment, docetaxel, in patients with pre-treated aNSCLC. However, these trials enrolled patients primarily from North America and Europe, and included very few Asian patients. It is noteworthy that the CheckMate 078 clinical trial, at the recent data cutoff (October 27, 2017), was the first Phase III trial to compare the efficacy and safety of nivolumab versus docetaxel in a predominantly Chinese population of patients (
N = 504) with pre-treated aNSCLC [
14], a statistically significant and clinically meaningful improvement in OS was found for patients assigned to nivolumab compared with docetaxel (12.0 vs 9.6 months, hazard ratio [HR] 0.68; 97.7% confidence interval [CI] 0.52–0.90;
p = 0.0006). The significant superiority with regard to OS was consistent across most prespecified subgroups, including histology and tumor PD-L1 expression level subgroups. Furthermore, nivolumab was well-tolerated, with a lower incidence of grade ≥ 3 treatment-related adverse events versus docetaxel (10% vs 48%).
Despite its remarkable clinical benefit and lower risk over conventional chemotherapy, nivolumab has induced high cost. Since 2017 the National Healthcare Security Administration (NHSA) has held a series of negotiations with pharmaceutical companies regarding oncology drugs, and many cancer drugs have been included in National Reimbursement Drug list (NRDL) with the price falling by an average of 56.7% [
15]. Unfortunately, due to the fact that nivolumab has not been approved by the China Food and Drug Administration (CFDA) until June 2018, it was not presented in those negotiations and little is known about its cost-effectiveness for Chinese NSCLC patients.
Therefore, the present study aimed to evaluate cost and effectiveness of nivolumab versus docetaxel for aNSCLC patients after prior systemic chemotherapy from the perspective of Chinese healthcare system. Information on the cost effectiveness of nivolumab is required by healthcare decision makers to determine the relative role of this novel treatments in aNSCLC.
4 Discussion
Nivolumab is the first PD-1 inhibitor proven to have a survival benefit for Chinese patients with aNSCLC, and was approved by China Food and Drug Administration (CFDA) in June 2018 [
14]. However, as this high priced ($60 per 4.5 mg) new drug has not yet been included in the NRDL, it is urgent to assess its value in terms of both efficacy and cost.
For patients with previously treated aNSCLC, the model-based costs and benefits (i.e. QALYs and LYs) for nivolumab arm were greater than docetaxel arm, the corresponding ICERs were ranged from $85,171 to 143,663 per QALY gained. Unfortunately, from the Chinese healthcare system perspective, nivolumab is unlikely to be cost-effective versus the standard second-line treatment docetaxel. The sensitivity analyses suggested a high likelihood of nivolumab being cost effective with increasing thresholds of WTP, which was consistent with previous study by Zeng et al. [
19]. Due to the unbalanced economic development in China, the per capita gross domestic product (GDP) among province-level administrative units differs significantly. For example, it ranged from $4735 in Gansu Province to $21,188 in Beijing city in 2018, therefore, the WTPs were set for general regions ($28,899, 3 × per-capita GDP in 2018) and affluent regions ($63,546, 3 × per-capita GDP of Beijing) in this study.
An important assumption made in discrete-time Markov models is that all state transitions occur simultaneously, at the end of each cycle. In reality, however, most kinds of transitions typically occur gradually throughout a time interval (on average, half-way through). This assumption does not affect reported probabilities (or the survival curve), but it may result in overestimation of expected survival in most models. For minimizing the influence on results, a half-cycle correction was adopted in current study.
The most influential parameters in our model were the price of nivolumab and patient weight, both of which were associated with the cost of nivolumab. From the perspective of cancer patients, the use of high-priced new drugs might impose a heavy financial burden, which likely leads to delay, abandonment, and discontinuation of treatment [
28]. According to the sensitivity analysis, we found that when the price of nivolumab was discounted by more than 30%, the ICER was lower than the WTP of $63,564/QALY, and nivolumab therapy began to be cost-effective in affluent regions. Fortunately, a number of oncology drugs have recently been included in the NRDL negotiation conducted by the Chinese government, with price decreases ranging from 30 to 70%. Therefore, negotiating nivolumab might be an effective way to make nivolumab less costly and more widely used in China.
It should be noted that, the utility of PFS played an important role in the model, and similar findings have been reported in previous studies [
19,
29]. A new study on the utilities of NSCLC from six countries including China conducted by Nafees et al, showed that some common toxicities associated with treatment probably lead to a significant decline in the utility of PFS [
20]. Toxicities including rash, fatigue, neutropenia and hair loss were observed in the CheckMate 078 trial [
14]. For higher accuracy, we weighted the utility of PFS according to the risks of these toxicities.
Other sensitivity parameters were found in our model, such as OS HR, and PFS HR, which were used to adjust the Weibull survival curve for the nivolumab arm. Moreover, we applied suitable HRs to determine whether the cost effectiveness of nivolumab varies in subgroups. The results indicated that nivolumab therapy for patients who were aged ≥ 65 years, female patients, and patients with ≥ 1% tumor PD-L1 expression resulted in lower ICERs ($85,171/QALY, $85,273/QALY and $90,309/QALY), which were strikingly consistent with previous study by Wan et al. [
30]. Thus, from a more far-sighted perspective, age stratification, tumor typing and PD-L1 assays before treatment would be helpful to use nivolumab more cost-effectively.
To our knowledge, there were only two published economic evaluations of nivolumab versus docetaxel as second-line treatments for aNSCLC, one from an Australian healthcare system perspective and the other from a broader societal perspective in Canada [
31,
32]. Both studies found that nivolumab was associated with substantial survival benefits at incremental cost, which was confirmed by our study. In addition, there is a similar economic study (conference abstract) from the Chinese healthcare payer’s perspective. The major finding was that patients receive nivolumab resulting in an ICER of $35,663 and $25,118 per QALY compared with docetaxel in squamous and non-squamous aNSCLC, respectively [
33]. The estimation of ICERs in our study was higher than that in Hu et al’s study, which mostly resulted from the different source of clinical data. Moreover, the sensitivity analysis in Hu et al.’s study found that the drug costs were a key model parameter, which is completely consistent with our results. However, it is noteworthy that we have further discussed the current findings on the basis of the National Reimbursement Drug List (NRDL) negotiation, which may be of great benefit to healthcare decision-makers in China.
Although other ICIs, such as pembrolizumab and atezolizumab, have recently been approved by the US Food and Drug Administration (FDA) as second-line settings for NSCLC patients [
34], we have not evaluated these therapies because of the lack of robust head-to-head trials, and the fact that they are not yet available in China. Most notably, atezolizumab has been confirmed to have similar efficacy and applicability to nivolumab based on the results of the Phase 2 POPLAR trial [
34]. In view of the largest population of lung cancer patients in China, further cost-effectiveness analyses are urgently needed to compare these innovative therapies from the perspective of China’s healthcare system.
5 Limitations
There are several limitations in the present study. First, we acknowledge that the CheckMate 078 trial is a large randomized Phase III trial and precisely designed to provide the most exhaustive evidence to clinical date, Even more remarkably, it is the only trial that compared nivolumab with docetaxel in a predominantly Chinese patient population with pre-treated aNSCLC, but our model depends largely on the validity and generalizability of this trial, and any biases within the trial will inevitably affect our model. Second, the unit costs of healthcare were pooled across various sources. We conducted a series of sensitivity analyses by varying the costs within a wide range and found a minimal impact on the ICER for both arms. Third, due to the lack of sufficient data in the subgroups, the results of the subgroup analyses in our study should be carefully interpreted, given that the assumption of the consistency of clinical data between subgroups and overall patients. Fourth, on the basis of the visual fit and statistical goodness-of-fit, the long-term survival of nivolumab was extrapolated from the clinical data of the CheckMate 078 trial, which would likely overestimate OS in the long term. Although there was a slight numerical difference in costs and effectiveness between the long-term and 21.2 months, which was the maximum follow-up time in the CheckMate 078 trial, differences with regard to model result were not significant, nevertheless, when mature OS data are available, it will be essential to further confirm the current findings.