Study selection
The database search identified 423 studies records, of which 394 were excluded as duplicates or did not meet the inclusion criteria. A total of 29 full-text studies were screened, of which nine studies were excluded due to: metastasis of colorectal cancer (n = 7), metastasis of neuroendocrine tumours of hepatic origin (n = 1), and intrahepatic cholangiocarcinoma (n = 1). Twenty studies met the eligibility criteria. A flow diagram of records founds, screened, selected, and full-text studies evaluated is shown in Fig. 1.
Bibliographic selection based on the PRISMA criteria
Overview of the included studies
Eleven of the 20 studies (55{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) were full economic evaluations [20,21,22,23,24,25,26,27,28,29,30] and nine studies (45{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) were partial evaluations [31,32,33,34,35,36,37,38,39] (Table 1). Using the CHEERS checklist, the thirteen articles were of high quality (mean score of 94{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}), and seven abstracts/poster were of lower quality assessment (mean score of 56{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}), mainly because of the limited breadth of data.
Full economic evaluations (n = 11)
Characteristics of the included studies
Eleven publications were categorized as full economic evaluations (7 articles [20, 22, 23, 26, 28,29,30] and 4 congress communications [21, 24, 25, 27]). Seven were published from a European perspective [22,23,24,25,26, 28, 29] and four from the USA [20, 21, 27, 30]. The HCC population studied were mainly patients with HCC in the intermediate and advanced stages (8 of 11 publications: one BCLC-B [23], four BCLC-C [24, 25, 27, 30], and three grouped stages BCLC-B and BCLC-C [26, 28, 29]); one publication grouped early and intermediate stages [22], and two publications grouped all three stages (BCLC-A, B and C) [20, 21].
Regarding the type of microsphere evaluated, three publications did not specify the type of microsphere [21, 26, 27]; two studies referred to TheraSphere® [22, 24], two studies referred to SIR-Spheres® [25, 29], three studies referred to both types (TheraSphere® and SIR-Spheres®) [20, 23, 30], and one study reported the use of three types of microspheres, including QuiremSpheres® [28]. The main comparators were TACE [20,21,22,23] and sorafenib [24,25,26,27,28,29,30, 30], in addition to transarterial embolization (TAE) [22], TACE with doxorubicin-releasing particles (DEB-TACE) [22] and lenvatinib [28].
Regarding the pharmacoeconomic parameters, two of the eleven studies were CEA [20, 21], eight were ACU [22,23,24, 26,27,28,29,30], and one was a CMA [25]. Six of the eleven studies used a Markov modelling [22,23,24, 26, 27, 30], two studies utilized Monte-Carlo modelling [20, 21], two were survival-based models [28, 29], and one utilized decision trees modelling [28]. The cost minimisation study did not specify the type of model [25] used. The time horizon ranged from 5 years [20, 21, 30] to lifetime [23, 26, 27, 29]. The payer’s perspective predominated (10 of 11 publications), although one study focused on the social perspective [28]. The outcome measures included overall survival (OS), life month gained (LMG), life years gained (LYG), quality-adjusted life years (QALY), incremental cost-effectiveness ratios (ICERs), incremental cost-utility ratios (ICURs), willingness-to-pay (WTP), and incremental net monetary benefit (NMBs). The characteristics of the full economic evaluations are summarized in Table 2.
TARE versus TACE
TACE therapy was one of the comparators considered in four of the eleven studies [20,21,22,23]); two studies [20, 21] compared TARE with TACE, a third study [22] included TACE and two other comparators (TAE and DEB-TACE), and lastly publication reported TACE as part of a sequence of therapies (TARE, TACE and possibly sorafenib [TTS sequence] versus TARE plus sorafenib [TS sequence]) [23]. The stages of the evaluated patients were heterogeneous; early [20,21,22], intermediate [20,21,22,23], and advanced [20, 21] disease.
TARE versus TKI
Seven studies [24,25,26,27,28,29,30] used systemic therapy as a comparator; 6 studies [24,25,26,27, 29, 30] reported only sorafenib as a comparator, and one study [28] included lenvatinib. Additionally, these seven studies evaluated patients with the intermediate-advanced disease.
Results of the full economic evaluations
The costs and health outcomes reported in the eleven studies were heterogeneous (Table 3).
TARE versus TACE
Four studies reported higher costs (TARE versus TACE) [20,21,22], and this finding was independent of the patient’s BCLC-A, B, or C in three studies. The fourth publication presented a higher cost in TS sequence therapy than TTS sequence (47{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} of patients with sorafenib) in patients with the intermediate disease [23].
In one study, the health outcomes reported for patients in the intermediate stage showed a benefit of TARE over TACE in terms of LYG and QALY [22]. The study evaluated sequences of therapies, TTS (with optional sorafenib), and showed a greater incremental benefit than TS for LYG and QALYs [23]. Two studies [20, 21]) reported the benefits for TARE in the advanced stage (BCLC-C), with lower benefits compared to TACE in the early and intermediate stages.
The ICERs of TARE versus TACE presented monthly (LMG) [20] and annual costs (LYG) [22]. Additionally, two studies [22, 23] presented ICUR results (€/QALY), and one study did not present any ratios [21]. For the early and intermediate stages of the disease, one study (Manas et al. [22]) presented an ICER of £ 12,833/LYG (£, 2020) (12,291 $US PPP/LYG) and established the ICUR of TARE versus TACE at £ 17,279/QALY (£, 2020) (17,397 $US PPP/QALY), with a 76.5{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} probability of being profitable considering a cost-effectiveness threshold of £ 20,000/QALY (£, 2020). In the intermediate stage, one study evaluated two treatment sequences and reported that TTS (with sorafenib in 47{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} of patients), including TARE, was the dominant strategy (i.e., it offered greater effectiveness with lower associated cost). When compared to TS, an 83{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} probability of being efficient based on a threshold of € 50,000/QALY was estimated [23]. In the advanced stage, TARE was superior to TACE (ICER 8 $US PPP/LMG) when the intervention was evaluated in one lobe and obtained an ICER of $ 356/LMG ($, 2013) (399 $US PPP/LMG) when the two-lobe intervention was evaluated [20]. TARE was inferior (with lower effectiveness and higher associated cost) when used in the early and intermediate stages [20]. The second publication by Rostambeigi et al. [21] did not detail the calculation of ICERs.
TARE versus TKI
Six [24,25,26, 28,29,30] of the seven studies compared TARE with sorafenib in patients with intermediate-advanced stage and reported lower costs for TARE (differences between 1454 to 46,982 $US PPP). However, Parikh et al. [27] evaluated a similar group of patients and reported conflicting cost results, a difference attributable to the source of the clinical trial efficacy parameters.
The benefits for health outcomes were greater for TARE [24,25,26, 29] than sorafenib in four of the seven studies (maximum QALY gained was 0.540 in BCLC-B, 0.27 in BCLC-C, and 0.601 in both stages); two studies [27, 28] showed greater health benefits for sorafenib (maximum QALY gained was 0.09), and one study [30] reported differing results depending on the source of clinical efficacy.
For patients with advanced-stage, TARE therapy was considered superior to sorafenib in five [24,25,26, 29, 30] of the seven studies when the SARAH RCT clinical parameters were used [7] as the source of clinical efficacy. The remaining two studies [27, 28] reported sorafenib was superior to TARE in patients with intermediate-advanced stage.
Study quality reporting assessment
Included studies categorized as full economic evaluations were appraised for their quality: six of the eleven studies (55{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) [22, 23, 26, 28,29,30] had a high score when evaluated with the 24-item checklist (mean compliance: = 99{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}). Approximately, 27{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} (3 of 11) and 18{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} (2 of 11) of the studies had a moderate score (mean compliance: 66{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) [20, 25, 27] and a low score (mean compliance of 46{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) [21, 24], respectively.
Partial economic evaluations (n = 9)
Characteristics of the included studies
Nine publications were partial evaluations (6 articles [31, 34,35,36,37, 39] and 3 congress communications [32, 33, 38]). Six publications were from the European perspective [31, 33, 36,37,38,39]), two from the United States [34, 35], and one from the Canadian perspective [32]. The HCC population included patients with intermediate and advanced stages in seven of the nine studies [31,32,33, 36,37,38,39]; five studies [31, 32, 36, 37, 39] reported the inclusion of patients as BCLC-B or BCLC-C, and two studies defined the intermediate or advanced stage as unresectable HCC (Muszbek et al.) [33, 38]. Of the two remaining studies, one (Ray et al.) [34] described HCC in a way that can be assumed to correspond to an early BCLC-A stage (male patient 65 years old with unresectable solitary HCC of 3 cm isolated in 1 lobe, not suitable for transplantation), and the second study (Ljuboja et al.) [35] did not define the population.
Three of the nine studies evaluated SIR-Spheres® [31, 35, 39], one included TheraSphere® [32], three considered both TheraSphere® and SIR-Spheres® [36,37,38], and two did not specify the type of microsphere evaluated. The comparators were TACE [31, 32, 34, 35, 38], ablative therapy [34, 35] and systemic therapies (sorafenib [31, 33, 36, 37, 39] and lenvatinib [39]).
Regarding the time horizon, six studies were CA [31, 33,34,35,36, 38] and reported time horizons ranging from 1 month to 2 years. The remaining three studies were BIA [32, 37, 39] and reported time horizons ranging from 3 years to a lifetime horizon. The payer’s perspective was most frequently used (100{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}); with the exception of one study that considered the social perspective [38]. The HCC stages of the study population, the comparators, and the outcome measures considered in the partial economic evaluations are highlighted in Table 4.
TARE versus TACE
Treatment with TACE was considered as a comparator in five [31,32,33,34,35] of the nine studies. Four of five studies reported the stages of HCC (early [34], intermediate, and/or advanced stages [31,32,33]). In studies of intermediate-stage HCC, one study compared only TACE versus TARE [33], two studies [31, 32] included sorafenib in addition to TACE, and two studies [34, 35] reported including radiofrequency ablation (RFA).
TARE versus TKI
Four studies [36,37,38,39] used systemic therapy as a comparator: three [36,37,38] reported sorafenib as a comparator, while one [39] publication also included lenvatinib in the assessment. All four studies considered patients in the intermediate-advanced stage.
Results of the partial economic evaluations
The costs and health outcomes were heterogeneous, mainly due to the type of economic evaluation performed and the grouping of patients with the different stages of the disease. Aggregated data for intermediate and advanced stages (BCLC-B combined with BCLC-C) were reported in five studies [31, 32, 36, 37, 39]. Data differentiated by HCC stages was reported in three studies (BCLC-A [34], BCLC-B [33], and BCLC-C [38]), and one publication [35] did not report the stage of disease (Table 5).
TARE versus TACE
Four CAs [31, 33,34,35] and one BIA [32] compared TARE versus TACE. The CA studies mostly indicated higher treatment costs (range: 11,572–42,368 $US-PPP) with TARE than with TACE (range: 9577–35,855 $US PPP) treatments [31, 33,34,35], ablative therapy (range: 3790–11,135 $US PPP) [34, 35] or sorafenib (12,460 $US PPP) [31]. However, one study (Muszbek et al.) [33] reported similar costs for TARE and TACE regardless of whether the costs were obtained from the official source (the NHS) or via a micro-costing approach [40]. Furthermore, Colombo et al. [31] highlighted the omission of the costs of unplanned hospitalization and adverse events (AEs) from their assessment. However, Ray et al. [34] established that in the early stage (based on a hypothetical cohort of patients older than 65 years) TARE had lower costs than TACE in more than one-third of the simulations of the evaluated scenarios. The BIA [32] study found cost savings with TARE during 3 consecutive years (savings of 40,699; 64,454, and 82,437 $US PPP at years 1, 2, and 3, respectively) of evaluation in a simulated population of 200 patients in a Canadian hospital.
No health outcomes were reported in the five studies that compared TARE with TACE. However, Colombo et al. [31] evaluated the treatment patterns in four centres in Italy and found TACE as the treatment of choice for intermediate HCC and sorafenib as the most commonly used first-line treatment for advanced HCC.
TARE versus TKI
The cost comparisons of TARE versus TKI (2 CA [36, 38] and 2 BIA [37, 39]) reported dissimilar results for TARE in patients with intermediate and/or advanced-stage disease. The CA by Lucà et al. [36] reported significantly lower cost for TARE (18,096 $US PPP) than sorafenib subgroup (28,520 $US PPP). Besides, the CA by Muszbek et al. [38] identified significant changes in the clinical practices for the management of advanced HCC patients, showing a 54 to 79{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} decrease in monthly costs compared to previous surveys. The BIA published by Rognoni et al. [37] from the Italian Health perspective was estimated to save € 7 million with the progressive increase in the use of TARE (from 20 to 50{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) instead of sorafenib over 5 years. The second BIA (Pollock et al.) [39] evaluated TARE versus without TARE in four European countries (Spain, France, Italy, and the United Kingdom) and reported the use of TARE in Spain would generate a cost savings of 26.5{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} over a 3-year period.
Within the type of resources used, the pharmacological cost, the work-up, the number of procedures and the management of AEs were identified as cost drivers for TARE and TKIs. Only three [36, 37, 39] of the four studies provided health outcomes in the survival rates [36], the number of events (deaths or hospitalizations) avoided [37], incremental LYG [39], and the proportion of patients receiving treatment with curative intent [39]. The CA by Lucà et al. [36] estimated that TARE had significantly higher medium-term survival rates than sorafenib (TARE 64.1{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f} vs. sorafenib 24.3{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}; p = 0.012) after 2 years of follow-up of patients with intermediate-advanced HCC. The BIA by Rognoni et al. [37] reported a greater number of deaths avoided (2 and 14 deaths in 5 and 10 years, respectively) and fewer hospital admissions due to hepatic decompensation (32 hospitalizations avoided in 5 years) in the intermediate-advanced stage. The BIA by Pollock et al. [39] reported an incremental LYG of 0.009 with TARE (1.176 LYG) compared to sorafenib (1.168 LYG) and reported that 71 additional patients would benefit from treatment with curative intent over a 3-year period.
Study quality reporting assessment
Approximately six [31, 34,35,36,37, 39] of the nine studies (67{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) had a high score when evaluated with a 20-items checklist (mean compliance:93{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}). The remaining three studies (33{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) were rated as having a moderate quality (mean compliance: 62{06e2b7c0a14bb457bf7dac0d18d586fd3fa1b0260a99c679eb119ff03d90035f}) [32, 33, 38].