Background and aims
Hepatocellular carcinoma and selective internal radiation therapy
Hepatocellular carcinoma (HCC) is the sixth most common cancer worldwide and the third-leading cause of cancer-related deaths [1]. Treatment options for HCC vary depending on the stage at which treatment is initiated; in earlier stages, treatments with curative intent may be viable such as resection and liver transplant, while in later stages, options include locoregional therapies, systemic therapies or best supportive care, depending on the disease stage and liver function, among other factors [2]. After the publication of results from the IMbrave150 randomized controlled trial (RCT) in 2020 [3], treatment guidelines from the European Society of Medical Oncology (ESMO) were updated to recommend the use of atezolizumab–bevacizumab as first-line therapy in patients with Barcelona Clinical Liver Cancer (BCLC) stage C HCC, for whom atezolizumab–bevacizumab is not contraindicated, with sorafenib and lenvatinib recommended otherwise [2, 4]. In patients with liver-confined disease and good liver function (Child-Pugh class A), where systemic therapy is not feasible, ESMO recommend the use of selective internal radiation therapy (SIRT), a locoregional, intraarterial therapy that results in overall survival (OS) outcomes not significantly different from those in patients treated with sorafenib [5, 6].
Relationship between quality of life and adverse events
A key challenge in treating patients with more advanced disease, such as those with HCC in BCLC stage C, is that any detrimental effect of treatment on patient quality of life (QoL) is highly undesirable; even with maintained liver function and judicious use of sequential systemic therapies, median OS is relatively short and maintaining patients’ QoL is therefore of the utmost importance [7]. Indeed, an evaluation of treatment preferences in patients with advanced HCC concluded that patients were willing to trade some months of life to avoid risks of adverse events (AEs) and maintain their QoL [8]. Unfortunately, many treatments for advanced HCC are associated with high AE incidence rates. In the IMbrave150 safety population, for instance, grade 3 or 4 AEs (severe and life-threatening, respectively), occurred in 56.5% of 329 patients in the atezolizumab–bevacizumab group and in 55.1% of 156 patients in the sorafenib group [3]. In the placebo-controlled RCT of sorafenib that resulted in the adoption of sorafenib as the first-line standard of care for patients with advanced HCC, the overall incidence of treatment-related AEs was 80% in the sorafenib group versus 52% in the placebo group, with grade 3 (i.e., severe) drug-related diarrhea and hand-foot skin reactions occurring significantly more frequently in the sorafenib group [9]. The Sorafenib Versus Radioembolization in Advanced HCC (SARAH) study, a multicenter, phase 3, randomized, open-label trial evaluated the efficacy and safety of sorafenib versus SIRT using Y-90 resin microspheres in patients with locally advanced and inoperable HCC. The trial demonstrated there were fewer patients with at least one grade ≥ 3 treatment-related AEs in the SIRT group compared to the sorafenib group (41% versus 63% respectively).
Quality of life associated with selective internal radiation therapy and sorafenib in the SARAH trial
The SARAH trial evaluated patient QoL using the European Organisation for Research and Treatment of Cancer (EORTC) QOL questionnaire (QLQ-C30 version 3) and a specific HCC module, the QLQ-HCC18 [5]. Key baseline characteristics and QoL scores for the intent-to-treat population in the SARAH trial are presented in Table 1 [5, 10]. The global health status (GHS) scale scores from the QLQ-C30 were significantly better in the SIRT arm than in the arm treated with sorafenib (p = 0.0048), and the difference between the treatment arms increased with time (group-time interaction p = 0.0447), with the authors concluding that QoL and tolerance might therefore help when choosing between the two treatments [5]. These results were aligned with a further analysis from the SARAH trial confirming that the time-to-deterioration (TTD) in QoL was significantly longer with SIRT versus sorafenib, with the authors stressing the role of these results in optimizing the clinical benefit to patients [10]. Moreover, a recent indirect comparison of SIRT with atezolizumab-bevacizumab showed that both treatments may achieve similar TTD in QoL, thereby enriching the evidence base that informs patient and clinician decision-making [11].
Table 1
Baseline characteristics and QoL scores for participants in the SARAH intent-to-treat population
Baseline characteristics [5] | SIRT (n = 237) | Sorafenib (n = 222) |
|---|---|---|
Median age, years (IQR) | 66 (60–72) | 65 (58–73) |
Gender (%) | ||
Male | 212 (89) | 202 (91.0) |
Female | 25 (11) | 20 (9.0) |
Mean BMI (kg/m2) | 27.1 (5.3) | 27.4 (4.5) |
ECOG performance status (%) | ||
0 | 145 (61) | 139 (63) |
1 | 92 (39) | 83 (37) |
BCLC stage (%) | ||
A | 9 (4) | 12 (5) |
B | 66 (28) | 61 (27) |
C | 162 (68) | 149 (67) |
Child-Pugh liver function class (%) | ||
A5 + A6 | 196 (83) | 187 (84) |
B7 | 39 (16) | 35 (16) |
Unknown | 2 (1) | 0 (0) |
Baseline EORTC QLQ-C30 questionnaire scores10 | SIRT (n = 122) | Sorafenib (n = 163) |
Mean Global health status (n) | 65.3 (110) | 65.1 (154) |
Mean Functional scales score (n) | ||
Physical | 83.4 (122) | 82.6 (163) |
Role | 83.6 (122) | 80.0 (163) |
Emotional | 77.4 (122) | 71.0 (163) |
Cognitive | 85.7 (122) | 84.8 (163) |
Social | 88.1 (122) | 84.9 (161) |
Mean Symptom scales/items score (n) | ||
Fatigue | 29.9 (122) | 32.9 (163) |
Nausea and vomiting | 4.5 (122) | 6.3 (163) |
Pain | 19.4 (122) | 20.8 (163) |
Dyspnea | 27.5 (120) | 23.9 (163) |
Insomnia | 27.0 (122) | 33.5 (161) |
Appetite loss | 18.9 (122) | 18.1 (162) |
Constipation | 13.9 (122) | 17.1 (162) |
Diarrhea | 12.7 (121) | 15.1 (161) |
Financial difficulties | 12.8 (122) | 10.3 (162) |
Given the importance of understanding and preserving QoL in patients with advanced HCC, the objective of the present study was to identify the AEs contributing to changes in QoL in patients with unresectable HCC using data from the SARAH trial.
Methods
The SARAH trial
Patient-level data on QoL and incidence of AEs were obtained from the SARAH trial [5]. The main inclusion criteria for the SARAH trial required patients to be at least 18 years old; have a life expectancy greater than 3 months; a diagnosis of either locally advanced HCC (BCLC stage C) with no extrahepatic disease, new HCC that was ineligible for surgical resection, liver transplantation, or thermal ablation following previously cured hepatocellular carcinoma, or HCC that failed to respond to two unsuccessful rounds of transarterial chemoembolization (TACE) [5]. Patients were also required to have an Eastern Cooperative Oncology Group (ECOG) performance status score of either 0 or 1 and a Child-Pugh liver function class A or B score less than or equal to seven [5].
Patients were excluded from the study if they had extrahepatic metastasis or an additional primary tumor (excluding basal cell carcinoma or superficial bladder cancer). Patients who had received previous treatment for the current nodule (excluding TACE), or had active gastrointestinal bleeding, encephalopathy, or refractory ascites were also excluded. Those with a contraindication to hepatic embolization were also ineligible for inclusion within the study. Full details of the inclusion and exclusion criteria have been reported previously by Vilgrain et al. [5].
Descriptive analysis of changes in QoL
First, descriptive analyses of the data were conducted, summarizing the frequency of all AEs and the proportions of patients reporting each AE reported in the SARAH trial. Data from the EORTC QLQ-C30 questionnaire were then used as the basis of the QoL analyses [13]. The QLQ-C30 consists of 30 questions in total and includes a mix of multi-item scales and single-item measures as follows: five functional scales (physical, role, emotional, cognitive, and social functioning); three symptom scales (fatigue, pain, and nausea/vomiting); six single questions (assessing dyspnea, appetite loss, sleep disturbance, constipation, diarrhea, and the perceived financial impact of disease and treatment). The instrument also includes a GHS scale. QLQ-C30 measurements were available at baseline, 1 month, and every 3 months thereafter, from patients’ hospital visits [5]. For the purposes of the present study, the physical, role, emotional, cognitive, and social functioning scales and GHS scores were calculated in line with the linear transformations detailed in the QLQ-C30 scoring guide.
Changes in the GHS score, physical, role, emotional, cognitive, and social functioning scales from the previous visit and frequencies of all AEs happening between each study visit were first calculated. Where QLQ-C30 were missing, no imputation was performed; patients with no baseline or follow-up QoL assessments were excluded from the analysis. Two separate approaches were then employed to analyze the contributions of different AEs to changes in QoL: one in which the AEs were combined into groups as specified in the SARAH statistical analysis plan, and one using principal component analysis (PCA) to select the events for analysis. Given the large number of different AEs experienced in the SARAH trial, the PCA was conducted to reduce the dimensionality of the data, i.e., the number of potential predictor variables, while maximizing the retained variability [14].
PCA of adverse events
The PCA was run using the prcomp function from the R stats package (version 4.1.1) on a matrix of the observations of each AE, with events grouped by severity based on Common Terminology Criteria for Adverse Events (CTCAE) grades; specifically, events were pooled into one of two groups: CTCAE grades 1 or 2 (mild or moderate), and CTCAE grades 3–5 (severe, life-threatening, or fatal). Based on visual examination of the scree plot from the PCA, the 20 largest contributors to dimension 1 of the PCA were retained, and the 10 largest contributors to dimensions 2 and 3. The relevance of AEs identified with the PCA was confirmed by clinician consultation, and AEs not captured by the PCA but deemed important by clinicians were added to the list of selected AEs. Correlations between the incidence of AEs selected in the PCA were assessed by calculating Pearson correlation coefficients between each AE using the cor function in the R stats package.
Quantitative analysis of the relationship between AEs and QoL scores
Linear mixed-effects models were then used to analyze the association between key AEs and the change in GHS score and the physical, role, emotional, cognitive, and social functioning scales, for all available time points for a patient versus the preceding time point. Models were fitted using the lmer function in the lme4 R package by minimizing the restricted maximum likelihood criterion [15]. In addition to the AEs selected, two additional covariates were included in the models based on previous studies: age and ECOG performance status [11, 16]. The study on ECOG performance status showed that ECOG may be an effect modifier of TTD in QoL at the 10% significance level. Any potential differential effect of treatment (SIRT or sorafenib) on QoL was assessed by also evaluating one model including a random intercept for treatment arm; goodness-of-fit of the models with and without the random treatment intercept were then evaluated using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) [17, 18]. Subsequent linear mixed-effects models were then developed capturing only those events that fell below a significance threshold of p < 0.20, and compared with both the full models using the AIC and BIC. In the final models, a significance threshold of 0.1 was adopted and significance of each model parameter was evaluated using t-tests based on Satterthwaite’s method [19, 20].
Results
The analysis was conducted on 431 participants from the SARAH trial, compared to 442 in the SARAH safety population (Table 1) [5]. The difference of eleven participants (2.5% of the original safety population) was the result of removing patients for whom both a baseline and at least one follow-up QLQ-C30 measurement were not available.
Grouping AEs
The analysis in which AEs were grouped in alignment with the SARAH statistical analysis plan showed that grade 1–2 and grade 3–5 gastrointestinal disorders (comprising anorexia, diarrhea, nausea or vomiting, abdominal pain, gastrointestinal ulceration and gastrointestinal bleeding), grade 3–5 liver disorders (comprising ascites, liver dysfunction, and radiation hepatitis), and grade 3–5 constitutional symptoms (comprising infection, fever, fatigue, and weight loss) were the only significant event groups in the mixed linear model of the GHS scale (Fig. 1).
Fig. 1
Parameter estimates from the linear mixed model of adverse event groups as drivers of the QLQ-C30 global health status scale score based on adverse event grouping as per the SARAH statistical analysis plan. ECOG, Eastern Cooperative Oncology Group; QoL, quality of life; n denotes the total number of patients experiencing at least one event in the adverse event category across both the sorafenib and SIR-Spheres Y-90 resin microspheres arms of the SARAH trial. Mean values < 0 plotted in red, mean values ≥ 0 plotted in blue
×
![]()
Selection of AEs for the analysis
Based on the results of the PCA, a total of 28 AEs were ultimately included in the mixed models (Table 2, Supplementary Fig. S1, and Supplementary Tables S1–S2); all 20 AEs from dimension 1 of the PCA and 8 unique AEs from dimension 2 were included. All AEs included in dimension 3 of the PCA had already been included in dimensions 1 and 2, resulting in 28 AEs for inclusion in the models. The first three components explained 28% of the total variance.
Table 2
Adverse events and common terminology criteria for adverse events grades selected for inclusion in the linear mixed-effects models
Adverse event | Grade |
|---|---|
Anemia | 1–2 |
Constitutional symptoms | |
Appetite decreased | 1–2 |
Asthenia | 1–2 |
Fatigue | 1–2 |
Pyrexia | 1–2 |
Weight Decreased | 1–2 |
Dermatological events | |
Palmar Plantar Erythrodysesthesia Syndrome | 1–2 |
Gastrointestinal disorders | |
Abdominal pain | 1–2 |
Diarrhea | 1–2 |
Nausea | 1–2 |
Hypertension | 1–2 |
Laboratory anomalies | |
Alanine Aminotransferase Increased | 1–2, 3–5 |
Aspartate Aminotransferase Increased | 1–2, 3–5 |
Blood Alkaline Phosphatase Increased | 1–2 |
Blood Bilirubin Increased | 1–2 |
Blood Creatinine Decreased | 1–2 |
Hyperbilirubinemia | 1–2 |
Hyperkalemia | 1–2 |
Hypoalbuminemia | 1–2 |
Hypokalemia | 1–2 |
Hyponatremia | 1–2 |
Leukopenia | 1–2 |
Lymphopenia | 1–2, 3–5 |
Neutropenia | 1–2 |
Thrombocytopenia | 1–2 |
Based on the Pearson correlation coefficients, none of the events were correlated to an extent that warranted removal from the analysis – a correlation of 0.51 was observed for grade 1–2 aspartate and alanine aminotransferase elevation while all other absolute correlation values were < 0.45 (Supplementary Figs. S2 and S3). Clinical expert opinion confirmed these AEs as relevant and suggested no additional AEs for inclusion. Three of the selected events were CTCAE grades 3–5, with the remaining events all grades 1–2. The most common events were grade 1–2 diarrhea, fatigue, and elevated aspartate aminotransferase (AST), occurring in 30.6%, 29.9%, and 27.6% of patients overall (Supplementary Table S1).
Linear mixed-effects models for QoL scales
The AIC and BIC criteria showed that linear mixed-effects model omitting the random intercept for treatment arm were generally the best fit and these models were therefore used for the full analyses. In each case, the subsequent models including AEs selected using a p < 0.20 threshold then consistently outperformed the models that omitted the random intercept and were therefore used for the final analyses. Parameters are presented for the full models prior to selection of significant variables in the supplementary materials (Supplementary Tables S2–S7).
The model for the GHS scale showed that diarrhea, decreased appetite, abdominal pain, and palmar-plantar erythrodysesthesia syndrome (hand-foot syndrome) were the only statistically significant parameters at a 5% significance level, all with negative effects on the GHS scale score (Fig. 2). The largest parameter was abdominal pain with a coefficient value of -6.10 (standard error [SE] of 1.94; p = 0.002) on the underlying scale score range of 0–100.
Fig. 2
Parameter estimates from the linear mixed model of the 7 adverse events selected by principal component analysis and remaining as significant drivers of quality of life at the p < 0.20 level measured based on the QLQ-C30 global health status scale score. QoL, quality of life; n denotes the total number of patients for whom global health status change was available and who experienced at least one event in the adverse event category across both the sorafenib and SIR-Spheres Y-90 resin microspheres arms of the SARAH trial. Mean values < 0 plotted in red, mean values ≥ 0 plotted in blue
×
![]()
Only the anemia, fatigue, and abdominal pain parameters were significant in the physical function scale model (Supplementary Fig. S4), with the largest parameter being anemia with a coefficient value of -5.33 (SE 1.23; p < 0.001). Elevated AST, diarrhea, and nausea were the only significant parameters in the role function model, with nausea having the largest parameter estimate of those at -5.40 (SE 2.61; p = 0.038; Supplementary Fig. S5). Anemia, fatigue, neutropenia, hyperbilirubinemia, and elevated aspartate aminotransferase (AST) all featured across the emotional (Supplementary Fig. S6), cognitive (Supplementary Fig. S7), and social functioning (Supplementary Fig. S8).
Discussion and conclusions
The analysis of AEs grouped into the pre-specified categories defined in the SARAH statistical analysis plan showed that grade 1–2 and grade 3–5 gastrointestinal disorders, grade 3–5 liver disorders and grade 3–5 constitutional symptoms were the only significant drivers of reduced QoL as measured on the GHS scale of the QLQ-C30. Laboratory abnormalities were not found to be significant drivers of changes in QoL despite their high incidence (n = 326 patients with grade 1–2 events and n = 128 patients with grade 3–5 events across both arms of the SARAH trial). These findings confirmed the face validity of the modelling approach as laboratory test abnormalities would not necessarily manifest clinically and, at least at lower grades, would likely be asymptomatic and therefore undetectable to the patient as a possible driver of changes in their QoL.
The analysis of individual AEs selected by PCA showed that diarrhea, decreased appetite, abdominal pain, and hand-foot syndrome were significant drivers of reductions in QoL, highlighting that these events have a meaningful detrimental effect on general QoL. Anemia, fatigue and abdominal pain were found to significantly affect scores from the physical functioning scale, while elevated AST, diarrhea, and nausea were found to significantly affect scores from the role functioning scale.
Strengths and limitations
The key strengths of the present analysis were the use of a robust methodology to select the most relevant AEs for inclusion in the analysis, the use of linear mixed-effects models to establish the significance of the effects, and the reasonably large sample size of 442 patients from the SARAH trial. Furthermore, the PCA conducted to select the AEs yielded a list of AEs that broadly overlapped with the “events of interest” reported in the SARAH trial publication, with the most common events such as fatigue, diarrhea, hand-foot syndrome, abdominal pain, nausea, and liver function abnormalities included in both sets of AEs.
One of the primary limitations of the present analysis was that the AEs analyzed occurred in a single RCT of two treatments; sorafenib and SIR-Spheres Y-90 resin microspheres. The nature of the AEs occurring in the SARAH trial may therefore have been materially different from the AEs occurring in trials of other treatments and other treatment modalities; however, in the safety population of the SARAH trial, at least one serious AE occurred in 174 of 226 patients (77%) in the SIRT group and in 176 of 216 (82%) in the sorafenib group, and the events occurring in the sorafenib group might be expected to be similar to AEs in patients treated with other kinase inhibitors. A second potential limitation was the exclusion of patients with no baseline or follow-up QoL measurements. This approach may have excluded patients with the most severe disease and/or those experiencing the lowest QoL due to the incidence of severe AEs. This limitation has been noted in previous analyses of QoL data from the SARAH trial [10]; in the present study, this may have resulted in severe AEs being underrepresented in the final QoL analyses or the width of the confidence intervals around parameter estimates for severe AEs being inflated. The PCA also captured only 28% of the total variance, which might be considered a further limitation, but the resulting list of AEs was aligned with the key AEs reported in the SARAH trial.
Key drivers of QoL outcomes
Across the QoL scales evaluated, diarrhea was identified as a significant driver of QoL in the linear mixed-effects models of five of the six scales. The significance across these five scales shows the extent of the effect on patient QoL. While diarrhea is included as a single question and single-item symptom scale in the QLQ-C30 (“Have you had diarrhea?” with a recall period of one week), the response is not captured in any of the functional scales, highlighting further the detrimental effect of diarrhea on the ability of patients to conduct day-to-day physical tasks in addition to limiting the ability to pursue work, leisure, and other daily activities. The significant effects of decreased appetite, abdominal pain, and hand-foot syndrome on the GHS scale score were perhaps less surprising given the highly tangible nature of the AEs.
The finding of significance of some laboratory measures in the functional scale score models was unexpected, particularly grade 1–2 anemia (significant in the physical and emotional functional scales) and elevated AST (significant in the role, cognitive, and social functional scales). The most likely explanation for the effect of these laboratory measures on QoL is that the effects are mediated through symptomatic manifestations of the underlying laboratory value abnormalities, although this is challenging to demonstrate definitively. Notably, parameter estimates for grade 3–5 elevated AST were positive in both the social function and cognitive function models.
The CTCAE definition for anemia specifies hemoglobin levels between 10 g/dL and the lower limit of normal for grade 1 anemia, or levels of 8–10 g/dL for grade 2 anemia. Anemia is known to be a cause of fatigue in patients with cancer, and, while fatigue was also a significant driver of the physical function scale score model, the correlation between the AEs was negligible, with a Pearson correlation coefficient of just − 0.002 suggesting that the effects on QoL were mediated independently [21].
From the clinician’s perspective, the aim of treatment should be reducing the incidence and/or severity of the AEs experienced while on treatment or minimize their effect on patient QoL [22, 23]. The courses of action that may be taken to achieve these goals are diverse given the diverse array of AEs experienced, but in each case a combination of patient education, close monitoring, proactive management, and use of symptom-relieving medications can all help to reduce the impact of the AEs on QoL [20, 21]. However, as noted in the conclusions of the SARAH RCT, selection of the HCC treatment can also be a key determining factor in patient QoL, with the AE profiles of different treatments having a significant and also potentially clinically meaningful effect on overall patient QoL. For example, the AE profile of SIRT is favorable, e.g., when compared with sorafenib or TACE, and its lack of ischemic effects allows its use even when patients are not good candidates for TACE [5, 6, 24‐26]. The safety of SIR-Spheres Y-90 resin microspheres specifically was confirmed by Helmberger et al. [27], who documented 30-day mortality and serious AE rates of 1.0 and < 2.5%, respectively, across multiple hepatic malignancies.
The manifest AEs with the largest significant impact on the GHS scale were diarrhea, anemia, abdominal pain, and hand-foot syndrome. In the SARAH trial, the incidence of diarrhea, hand-foot syndrome, and abdominal pain were all notably higher in the sorafenib group than the SIRT group (12% vs. 63% for diarrhea; 0% versus 17% for hand-foot syndrome, and 19% versus 26% for abdominal pain). Establishing that these events have a significant impact on QoL in addition to occurring more frequently with sorafenib is an important finding and illustrates that sorafenib would be expected to have a significantly higher detrimental effect on QoL than SIRT. Given that other systemic therapies have been shown to have an AE profile similar to sorafenib, the conclusions of the present analysis may also apply to comparisons of SIRT with other systemic therapies [3, 28].
Understanding the drivers of QoL in patients with HCC is essential in the context of patient centricity and the role of the patient in decision-making. The QLQ-C30 data used in the present analysis are patient-reported and are therefore critical in understanding how patients feel their treatment and, in particular, treatment-related AEs affect their QoL. There may be a mismatch between patient perceptions and clinicians’ understanding of the impact of treatment safety profile, and one of the primary objectives of the present study was to quantify this to help clinicians better understand the value that patients may place on treatments. Finally, understanding the QoL drivers is also critical from the health economic and policymaker perspective. A recent study has shown that modelling data from different trial populations within the same trial (e.g., safety versus intent-to-treat) and using different modelling techniques can yield directionally opposed estimates of quality-adjusted life expectancy, even when modelling data from a single RCT [29]. Understanding the clinical events that drive underlying changes in QoL is therefore essential in this context, and the present analysis can support clinician and patient discussion to develop decision-making according to what is important for the patient.
When deciding on using or reimbursing a treatment, clinicians, patients, and decision-makers should consider treatment profiles beyond AEs. Availability, for example, is more limited for SIRT relative to sorafenib as SIRT requires a thorough work-up and must be performed in specialist centers [6, 23]. Consideration of costs and cost-effectiveness, in contrast, would, in the choice between sorafenib and SIRT with Y-90 resin microspheres, favor Y-90 resin microspheres as it has been shown to be a potentially dominant, i.e., more efficacious and less costly, alternative to sorafenib in the UK and to be associated with cost savings across multiple settings, including Brazil, France, Italy, Spain, and the UK [30‐32].
Conclusions
In conclusion, the present analysis indicated that a number of tangible AEs, particularly diarrhea, hand-foot syndrome, abdominal pain, and nausea, have a significant negative impact on QoL in patients with unresectable HCC and, given the typically poor prognosis, this should be taken into full consideration when selecting treatments for HCC in this patient group.
Acknowledgements
This study was funded by Sirtex Medical United Kingdom Ltd. The authors thank Waqas Ahmed and Johannes Pöhlmann (Covalence Research Ltd) for their assistance with preparing this manuscript.
Declarations
Ethics approval
The study was conducted in accordance with the Declaration of Helsinki, approved by an ethics committee, and complied with the provisions of the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use Guidelines for Good Clinical Practice.
Consent to participate
Informed consent was obtained from all individual participants included in the study.
Competing interests
IA was a full-time employee of Sirtex Medical Europe GmbH at the time of the study. RFP is a director, shareholder, and full-time employee of Covalence Research Ltd, which received consultancy fees from Sirtex Medical United Kingdom Ltd to run the analyses, generate tables and figures, and draft the manuscript. PLC was a full-time employee of Sirtex Medical Europe GmbH at the time of the study. VKB was a full-time employee of Sirtex Medical United Kingdom Ltd at the time of the study. SS is a full-time employee and director of Sirtex Medical United Kingdom Ltd. VV declares research funding and speaker fees from Sirtex. HP and GC have no conflicts of interest to declare.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
