If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author at: Division of Cancer Care and Epidemiology, Cancer Research Institute at Queen’s University, 10 Stuart Street, 2nd Level, Kingston, Ontario K7L3N6 Canada.
Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Ingham Institute for Applied Medical Research, University of New South Wales (UNSW), Liverpool, AustraliaDivision of Cancer Care and Epidemiology, Cancer Research Institute at Queen’s University, Kingston, Canada
Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Ingham Institute for Applied Medical Research, University of New South Wales (UNSW), Liverpool, AustraliaSouth Western Sydney Clinical School, UNSW, Sydney, Australia
Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Ingham Institute for Applied Medical Research, University of New South Wales (UNSW), Liverpool, Australia
Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Ingham Institute for Applied Medical Research, University of New South Wales (UNSW), Liverpool, AustraliaDepartment of Radiation Oncology, Prince of Wales Hospital, Sydney, Australia
Collaboration for Cancer Outcomes Research and Evaluation (CCORE), Ingham Institute for Applied Medical Research, University of New South Wales (UNSW), Liverpool, Australia
To describe the population benefit of radiotherapy in a high-income setting if evidence-based guidelines were routinely followed.
Methods
Australian decision tree models were utilized. Radiotherapy alone (RT) benefit was defined as the absolute proportional benefit of radiotherapy compared with no treatment for radical indications, and of radiotherapy over surgery alone for adjuvant indications. Chemoradiotherapy (CRT) benefit was the absolute incremental benefit of concurrent chemoradiotherapy over RT. Five-year local control (LC) and overall survival (OS) benefits were measured. Citation databases were systematically queried for benefit data. Meta-analysis and sensitivity analysis were performed.
Findings
48% of all cancer patients have indications for radiotherapy, 34% curative and 14% palliative. RT provides 5-year LC benefit in 10.4% of all cancer patients (95% Confidence Interval 9.3, 11.8) and 5-year OS benefit in 2.4% (2.1, 2.7). CRT provides 5-year LC benefit in an additional 0.6% of all cancer patients (0.5, 0.6), and 5-year OS benefit for an additional 0.3% (0.2, 0.4). RT benefit was greatest for head and neck (LC 32%, OS 16%), and cervix (LC 33%, OS 18%). CRT LC benefit was greatest for rectum (6%) and OS for cervix (3%) and brain (3%). Sensitivity analysis confirmed a robust model.
Interpretation
Radiotherapy provides significant 5-year LC and OS benefits as part of evidence-based cancer care. CRT provides modest additional benefits.
]. The population-level benefits of evidence-based use of radiotherapy in high-income countries have been estimated for specific cancers by using a model-based approach, although the benefit of radiotherapy to the overall cancer population has not yet been described in this way [
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
]. Such information would be useful for informing health policy, quality improvement, and for performing economic analyses of radiotherapy.
In this report, the proportion of the whole cancer population deriving 5-year local control and overall survival benefit from radiotherapy is described. The population benefits of radiotherapy alone, and the additional benefit of concurrent chemotherapy with radiotherapy were estimated. The benefit to the subset with curative indications is described. The impact of sources of uncertainty on model estimates was quantified.
Methods
Defining indications for radiotherapy
A previously described population-based decision tree model was used to measure the proportion of patients with each evidence-based indication for radiotherapy (RUR) in the cancer population of Australia [
Barton MB, Jacob S, Shafiq J, Wong KH, Thompson S, Delaney G, et al. Review of optimal radiotherapy utilisation rates. http://tinyurl.com/pwkua34. Sydney: Ingham Institute, Collaboration for Cancer Outcomes Research and Evaluation; 2012.
]. TreeAge Pro 2008 (Release 1.6, TreeAge Software, Inc.) was utilized to build, depict and analyze the model. Evidence-based indications in favor of first-course radiotherapy were based upon superior local control, toxicity profile, quality of life and/or overall survival. Indications were identified based on evidence-based treatment guidelines from national and international organizations [
]. The highest level of epidemiological evidence was utilized, according to a pre-specified hierarchy, in order to define the population-based proportion of patients with patient-related and disease-related characteristics defining the incidence of each radiotherapy indication [
]. Western population data were used to estimate the incidence of each indication in Australia. Australian epidemiological data were used where available. In clinical situations where radiotherapy was considered an equal option with surgery or chemotherapy, the other options were included in the model and sensitivity analysis was undertaken to determine a possible range in population proportion that radiotherapy was indicated for. In specific cases, the radiotherapy benefit model was expanded to account for end nodes in the decision tree where there were subgroups with different benefits of radiotherapy such as groupings of age, performance status or presence of different RT indications. In these cases, the methods used to develop the original model were utilized [
Population benefit was defined as the absolute proportion of patients in the overall cancer population that benefited from external beam radiotherapy.
Benefit to the treated population was defined as the benefit of curative radiotherapy (i.e. radical or adjuvant) among patients with curative radiotherapy indications.
Endpoints were 5-year overall survival (OS) and 5-year local control (LC). These provided measures of radiotherapy benefit taking into account competing risks (OS), and a measure of the local effects of radiotherapy in the absence of competing risks (LC). For prostate cancer, LC was conservatively estimated as equivalent to biochemical control. Palliative benefits and brachytherapy alone benefits were not considered.
Radiotherapy alone (RT) benefit was estimated separately from the additional incremental benefit of concurrent chemotherapy and radiation (CRT). Radical RT benefits were defined as the absolute proportional benefit of RT over no treatment, and for adjuvant or neoadjuvant RT, the absolute proportional benefit of radiotherapy plus surgery over surgery alone. CRT benefit was the absolute proportional benefit of concurrent chemotherapy and RT over RT alone.
Systematic review of evidence of radiotherapy benefit
Systematic review was undertaken to define the highest level of clinical evidence defining the benefit (LC or OS) for each radiotherapy indication. The Australian National Health and Medical Research Council hierarchy of evidence [
National Health and Medical Research Council. NHMRC levels of evidence and grades for recommendations for developers of guidelines: Australian Government; 2009.
] was used to rank evidence. Searches were undertaken in Ovid, querying Medline, Embase, and all evidence-based medicine sources (including Cochrane CENTRAL). This provided a comprehensive basis from which to identify studies reflecting outcomes of treatment in a high-income setting, including abstract-only sources. To supplement these queries, publicly available population-based outcome data from SEER were queried. To ensure completeness, hand searches of key article reference lists were performed, and Pubmed and Google Scholar were queried using keywords and related article searches. Prior publications and reports provide example search strategies [
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
]. In cases where more than one source of the same evidence level was identified, meta-analyses were performed. Generic inverse variance meta-analysis was performed using Review Manager software (Version 5.1–5.3, The Nordic Cochrane Centre, The Cochrane Collaboration). All searches for radiotherapy benefit were completed between January 2012 and June 2016. Guidelines were reviewed to ensure radiotherapy indications were up to date, up to at least June 2015. Results published in peer-reviewed manuscripts supersede earlier reports.
Radiotherapy population benefit
Radiotherapy population benefit was determined by multiplying the absolute proportional benefit of each radiotherapy indication by the absolute proportion in the whole cancer population with the indication, and then summing all such products. For example, in the simplified model of glottis cancer population 5-year OS benefit depicted in Fig. 1, there are two indication benefits (stage I–II radiotherapy, and stage III–IVB radiotherapy). The population benefit was calculated as: (% of stage I–II RT patients with treatment benefit) × (proportion of all glottis cancer with stage I–II RT indication) + (% of stage III–IVB RT patients with treatment benefit) × (proportion of all glottis cancer with stage III–IVB RT indication) = 0.62 × 0.66 + 0.20 × 0.20 = 0.45. This means that for this model, 45% of all glottis cancer patients would survive to 5 years due to radiotherapy utilized according to guidelines, as compared to no use of radiotherapy.
Fig. 1Simplified radiotherapy population benefit model for glottic larynx cancer 5-year radiotherapy alone overall survival.
Deterministic (univariate) and probabilistic (multivariate) sensitivity analyses were undertaken. TreeAge Pro 2008 software was utilized. Uncertainties considered were: [
] Uncertainty in the magnitude of radiotherapy benefit.
Deterministic sensitivity analysis is depicted in tables, showing multiple one-way sensitivity analyses ordered according to the magnitude of influence on the benefit estimates.
Probabilistic sensitivity analysis was performed using Monte Carlo (MC) simulation. Standard errors were defined for all benefit estimates, utilizing previously described formulae for extracting summary statistics from published manuscripts [
]. Flat probability distributions were utilized for epidemiological estimates where there was a range of values considered equally plausible. 10,000 iterations of each MC simulation were performed, with the 95% confidence interval defined based on the 2.5th and 97.5th percentile benefits.
Results
48% of all cancer patients had an indication for radiotherapy. 34% of all cancer patients had first course radical, adjuvant or neoadjuvant indications for radiotherapy and 14% had first course palliative indications. Radical, adjuvant and neoadjuvant indications represented 71% of all indications for first-course radiotherapy. 39% of all 170 curative radiotherapy indications were supported by level I or II evidence. The proportion supported by level I or II evidence was less for radiotherapy alone (25%) compared to chemoradiation (73%). The low proportion supported by level I/II evidence for radiotherapy alone related to the many radical indications for radiotherapy that have become entrenched standards of care such as in head and neck and cervix.
Radiotherapy population benefit for all cancers
In univariate analysis, the population benefit for all cancers combined was for 5-year LC: 10.9% RT, 0.6% CRT. For 5-year OS: 2.4% RT, 0.3% CRT. These benefits are separate from those attributable to other forms of cancer therapy. For 2013, compared to no use of radiotherapy (RT or CRT), radiotherapy used according to guidelines translated into 13571 more Australians with local control, and 3485 more people who are alive at 5 years.
Benefit to the treated population
The benefit of curative radiotherapy among the population of patients with curative indications was for 5-year LC: 31.9% RT, 1.8% CRT. For 5-year OS: 7.1% RT, 0.9% CRT.
Radiotherapy population benefit for specific cancers
Considerations for defining specific radiotherapy indication benefits have been documented in great detail in prior publications and reports [
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
]. In systematic review, studies defining radiotherapy indication benefit were most often excluded due to evidence level (e.g. exclusion of case series where clinical trial data existed), or insufficient stratification of outcomes data to define indication benefits.
Radiotherapy alone 5-year population LC benefit by cancer site was highest for cancers where radiotherapy was most commonly utilized for radical treatment: cervix 33%, head and neck 32%, prostate 26% (Fig. 2). Seven cancer sites had zero 5-year LC benefit: unknown primary, pancreas, ovary, liver, kidney, gallbladder and colon. Among all 26 cancer sites, the median LC benefit by site was 5%. The additional absolute population LC benefit of CRT ranged from 6% for rectum to nil for 18 cancer sites (Fig. 2).
Fig. 2Radiotherapy Population 5-year local control benefit. Ordered by magnitude of radiotherapy alone absolute proportional benefit. Radiotherapy benefit is separated into radiotherapy alone benefit and chemoradiation benefit.
Radiotherapy alone 5-year OS benefit by cancer site was highest for: cervix 18%, head and neck 16%, and vulva 11% (Fig. 3). Eight cancer sites had zero 5-year OS benefit, including melanoma. Among all 26 cancer sites, the median OS benefit by site was 1%. The additional absolute population OS benefit for CRT ranged from 3% for cervix and brain, to nil for 18 cancer sites (Fig. 3).
Fig. 3Radiotherapy Population 5-year Overall Survival Benefit. Ordered by magnitude of radiotherapy alone absolute proportional benefit. Radiotherapy benefit is separated into radiotherapy alone benefit and chemoradiation benefit.
For overall radiotherapy (RT plus CRT) 5-year LC benefit in Australia in 2013, the five sites with the most patients benefiting from radiotherapy were: prostate (4943), breast (2375), head and neck (1477), lung (1116), and rectum (787) (Table 1). For radiotherapy (RT plus CRT) 5-year OS benefit in Australia in 2013, the five sites with the most patients benefiting were: head and neck (778), lung (580), lymphoma (363), breast (353), and rectum (296) (Table 1).
Table 1Population number with 5-year LC and OS benefit from radiotherapy in Australia 2013. Ordered by number of Australians with OS benefit. 5-Year LC and OS benefit including RT and CRT benefit where indicated by evidence-based guidelines. Number of cancers in Australia is based on Australian Institute of Health and Welfare (AIHW) data. The remaining columns represent data derived from the model of radiotherapy benefit.
In deterministic sensitivity analysis, there were 134 elements of uncertainty in the model. Tables detailing the values and results of the multiple one-way sensitivity analyses are provided in Appendix 1. For 5-year LC population RT benefit, the maximal range of uncertainty was 10.1–12.3%. Four of the five greatest uncertainties were due to uncertainty in prostate cancer treatment preference. Of these, the most influential variable was the proportion of patients preferring external beam radiotherapy (EBRT) to radical prostatectomy (25–75% of patients). All other variables were far less influential in the RT 5-year LC model.
5-Year OS population benefit varied from 2.4% to 2.5%. Six of the 10 most influential uncertainties related to choice of surgery versus radiotherapy. The most influential factor in the RT 5-year OS model was the choice of surgery versus radical radiotherapy for HPV-related oropharynx cancer (24%–84% receiving radical radiotherapy). Other less influential variables included choice of surgery over radical radiotherapy for head and neck cancer of unknown primary, and laser versus radical radiotherapy for early glottic cancer.
CRT population benefit varied little. OS benefit varied from 0.33% to 0.36%, and LC benefit from 0.58% to 0.62%. For both CRT LC and OS benefit, the most influential factors related to uncertainty in the population-level data on patient selection for combined modality treatment. The most influential variable was uncertainty in the proportion of stage II–III bladder cancer patients receiving chemoradiation rather than cystectomy (0–24%). The next most influential variables were the proportion of non-small cell lung cancer patients with poor performance status (PS), unfit to receive concurrent chemoradiation (e.g. stage IIIB poor PS 12–38%).
In probabilistic sensitivity analysis of individual cancer sites, the model was robust (Table 2). The widest variation for individual cancer sites was for RT benefit for vulva cancer (e.g. LC 10% (95% confidence interval 2, 18)) and prostate cancer RT LC benefit (25% (19, 32)).
Table 2Summary of Monte Carlo simulation multivariate sensitivity analyses. Ordered by number of Australians with OS benefit as in Table 1.
Confidence intervals were not included for the total population effect, as the XRT and CRT models were distinct. The 95% confidence interval for the total population effect could thus not be directly estimated by a single run of Monte Carlo simulation that would account for correlated variables affecting both XRT and CRT benefit.
Confidence intervals were not included for the total population effect, as the XRT and CRT models were distinct. The 95% confidence interval for the total population effect could thus not be directly estimated by a single run of Monte Carlo simulation that would account for correlated variables affecting both XRT and CRT benefit.
Head and Neck
0.033
32% (28,36)
2% (1,2)
34%
18% (16,21)
2% (1,2)
20%
Lung
0.090
7% (5,8)
2% (1,2)
9%
4% (3,4)
1% (1,2)
5%
Lymphoma
0.042
11% (9,13)
0
11%
7% (5,9)
0
7%
Breast
0.122
15% (13,16)
0
15%
2% (1,4)
0
2%
Rectum
0.042
8% (6,10)
5% (4,6)
13%
3% (1,5)
1% (1,2)
4%
‘Other’ cancers
0.050
9% (9,9)
1% (0,1)
10%
3% (2,4)
0.2% (0,0.4)
3%
Prostate
0.185
25% (19, 32)
0
25%
1% (1,2)
0
1%
Cervix
0.008
32%(29,34)
4% (2,5)
36%
17%(15,18)
3%(1,5)
20%
Brain
0.014
8%(7,10)
1%(0,2)
9%
5%(4,5)
3%(1,5)
8%
Bladder
0.020
5% (3,7)
1% (0,1)
6%
3% (2,4)
1% (0,1)
4%
Testis
0.006
9% (8,11)
0
9%
9% (7,10)
0
9%
Uterus
0.020
6%(4,8)
0
6%
2%(1,3)
0
2%
Vulva
0.003
10% (2,18)
0
10%
8% (1,16)
0
8%
Esophagus
0.012
2% (1,3)
4% (1,6)
6%
0.3% (0,1)
2% (1,3)
2%
Stomach
0.018
2% (0,5)
0
2%
1% (0,4)
0
1%
Thyroid
0.018
1% (1,1)
0
1%
1% (1,1)
0
1%
Myeloma
0.012
2% (2,2)
0
2%
1% (1,1)
0
1%
Leukemia
0.023
0.4% (0.2,0.7)
0
0
1% (0,1)
0
1%
Colon
0.084
0
0
0
0
0
0
Gall Bladder
0.006
0
0
0
0
0
0
Kidney
0.023
0
0
0
0
0
0
Liver
0.012
0
0
0
0
0
0
Melanoma
0.099
3% (3,4)
0
3%
0
0
0
Ovary
0.013
0
0
0
0
0
0
Pancreas
0.021
0
0
0
0
0
0
Unknown Primary
0.024
0
0
0
0
0
0
All Cancer
1
10% (9,12)
1% (1,1)
11%
2% (2,3)
0.3% (0.2,0.4)
3%
* Confidence intervals were not included for the total population effect, as the XRT and CRT models were distinct. The 95% confidence interval for the total population effect could thus not be directly estimated by a single run of Monte Carlo simulation that would account for correlated variables affecting both XRT and CRT benefit.
In probabilistic sensitivity analysis for all sites combined, the model was robust. Estimates were for 5-year LC: RT 10.4% (9.3, 11.8), CRT 0.6% (0.5, 0.6). For 5-year OS: RT 2.4% (2.1, 2.7), CRT 0.3% (0.2, 0.4).
Discussion
We have estimated the benefit of radiotherapy to the population of cancer patients as a whole. Radiotherapy provides significant 5-year local control and overall survival benefits to cancer patients when utilized as part of a guideline-based cancer program. In a single year, compared to no use of radiotherapy, radiotherapy used according to guidelines can provide about 13,600 more Australians with local control, and 3500 more people with overall survival to 5 years. The vast majority of this benefit is derived from radiotherapy alone, and a modest proportion from the added benefit of concurrent chemotherapy. Among the treated population with curative indications, about one in three derived a 5-year LC benefit, and one in twelve a 5-year OS benefit.
When considering the magnitude of radiotherapy benefit that we have estimated, it is emphasized that the population radiotherapy benefits describe the absolute proportion of a total cancer population that derive a treatment benefit when radiotherapy is used according to guidelines, separate from the contribution of other modalities. Determining the number of patients benefiting from radiotherapy in a population is thus calculated as the population cancer incidence multiplied by the absolute proportional radiotherapy benefit. This differs from describing the average treatment benefit, or the pooled survival of all patients receiving radiotherapy. It is also noted that the absolute proportional benefit was affected by the approach taken for benefit estimation for cases where adjuvant radiotherapy was used; only the incremental benefit over surgery alone was attributed to radiotherapy. Most importantly, the benefits presented in this report represent an upper bound of benefit, when radiotherapy is optimally used according to guidelines for all patients with indications. Where access to curative radiotherapy is impaired, this will reduce the actual population benefit.
The magnitude of radiotherapy benefit was influenced by the proportion of all cases with adjuvant or radical indications, and the cancer incidence for each site. For example, although the population OS benefit for breast cancer was twelfth in order of magnitude among all cancer sites, due to the high breast cancer incidence this site was ranked fourth in terms of number of patients benefiting from radiotherapy (Table 1). Conversely, head and neck cancer incidence is low, but due to the large OS benefit of radiotherapy for this site it ranked first in number of patients with OS benefit. The magnitude of population benefit was also influenced by the proportion with curative indications that were adjuvant, given their smaller absolute benefit compared to radical indications. This was the case for breast cancer, given adjuvant indications predominated.
Some cancers had zero estimated five-year overall survival or local control benefit from radiotherapy used according to guidelines. This may be for a number of reasons. There may be benefits limited to time points earlier than five years. In the case of locally advanced pancreatic cancer, randomized trials suggest a possible benefit to chemoradiation at earlier time points, with any benefit at five years debatable due to poor long-term survival regardless of treatment [
Therapy of locally unresectable pancreatic carcinoma: a randomized comparison of high dose (6000 rads) radiation alone, moderate dose radiation (4000 rads + 5-fluorouracil), and high dose radiation + 5-fluorouracil: The Gastrointestinal Tumor Study Group.
A randomized phase III study of radiotherapy alone or with 5-fluorouracil and mitomycin-C in patients with locally advanced adenocarcinoma of the pancreas: Eastern Cooperative Oncology Group study E8282.
Length and quality of survival after external-beam radiotherapy with concurrent continuous 5-fluorouracil infusion for locally unresectable pancreatic cancer.
]. In other cases, there is insufficient clinical evidence to support routine use of radiotherapy as the standard of care. Colon cancer is an example of this. For some rare indications, such as for gallbladder cancer, radiotherapy is considered reasonable according to guidelines, but there is insufficient evidence to define benefit. Further details describing site- and indication-specific considerations defining benefit have been documented in publications and detailed online supplemental resources [
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
The magnitude of radiotherapy 5-year local control benefit relative to 5-year overall survival was driven by a number of factors. Generally the ranking of local control and overall survival benefit were similar (Fig. 2, Fig. 3), reflecting the contribution of local control to overall survival. The greatest exception was prostate cancer; the 5-year local control benefit (25.7%) was much greater than the 5-year overall survival benefit (1.1%). Radiotherapy obtains disease control at high rates for many prostate cancer indications, however, the only clinical trial-proven 5-year overall survival benefit of external beam radiotherapy that was identified in systematic review was for high-risk prostate cancer [
The population benefit model has a number of important strengths. It is robust, despite many model uncertainties. These related to uncertainties in optimal use of radiotherapy, and in the benefits of selected radiotherapy indications. These uncertainties were most often in smaller branches of the decision tree, limiting their impact on overall estimates. The approach taken is transparent, allowing peer review of inputs utilized to derive the incidence of each radiotherapy indication and its benefit. As mentioned, detailed information on the model is provided through publications and sources available online [
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
Barton MB, Jacob S, Shafiq J, Wong KH, Thompson S, Delaney G, et al. Review of optimal radiotherapy utilisation rates. http://tinyurl.com/pwkua34. Sydney: Ingham Institute, Collaboration for Cancer Outcomes Research and Evaluation; 2012.
].The model is readily adaptable, where relevant epidemiological data are available. It utilizes publicly available data sources.
As the largest and most influential branches of the model are determined by site-specific cancer incidence, basic estimates of benefit can be determined for nations with information on cancer incidence where further epidemiological data are unavailable. This approach was taken in the Lancet Commission report of the Global Task Force on Radiotherapy for Cancer Control (GTFRCC) to undertake an economic analysis of radiotherapy in Low- and Middle-Income Countries (LMIC) [
]. For the global cancer population, it was estimated that in 2012 alone, 1.5 million people would derive a local control benefit and more than 580,000 a survival benefit from optimal access to radiotherapy [
]. The annual number benefiting is projected to increase to 2.5 million (LC) and 950,000 (OS) by 2035.
Conversely, the consequences of inadequate investment in radiotherapy can be estimated. It has been estimated that among cancers considered by the GTFRCC model for 2015–35, 18 percent of lost economic output due to lack of optimal radiotherapy utilization in LMIC was for breast cancer [
]. In addition, the approach utilized to estimate radiotherapy benefit in the present report can be adapted to other modalities. For example, Do et al. are estimating the population overall survival benefit of chemotherapy [
Do V, Ng W, Delaney GP, Barton MB. An estimation of the population survival benefit of first-line chemotherapy for upper gastrointestinal cancer (#306) Clinical Oncology Society of Australia Annual Scientific Meeting; 2013.
There are limitations. Methodological issues in the source data are not eliminated by pooling the data. These include bias, confounding and generalizability. It is noted that the highest level of evidence from systematic review was always utilized for each indication. Indications with poor quality data were often less common indications, limiting their impact on the overall model. It is acknowledged that most radical radiotherapy indications represented entrenched practices because of obvious and irreplaceable benefit and had no randomized trial evidence against a control group receiving no active treatment. Given dismal historic outcomes for untreated cancer, there will never be equipoise sufficient to warrant such trials [
]. Additionally, model development is labor intensive. However, utilization of benefit estimates is far more rapid on model completion.
It is acknowledged that 5-year local control and 5-year overall survival, though very important, are not the only measures of population outcomes of radiotherapy or other forms of cancer treatment [
]. For example, there are palliative endpoints, including symptom control and quality of life. Also, for the most aggressive cancers where few patients survive to 5 years (e.g. pancreatic cancer), 2-year or shorter term outcomes are more appropriate. There were limitations in the availability of population-based epidemiological data required to estimate the incidence of symptoms and patient characteristics required to estimate the incidence of palliative indications, and also the incremental benefit of palliative radiotherapy over best supportive care alone.
The most influential uncertainties in the model are informative. The greatest uncertainties were for indications where other alternatives existed, for example early prostate cancer. Active surveillance, radical external beam radiotherapy, brachytherapy, and radical surgery are all options for patients suitable for potentially curative management. The model used for this study describes optimal radiotherapy utilization according to guidelines. However, modeling in preference-sensitive situations such as prostate cancer treatment requires accurate information on patient preference. These data are limited and incomplete [
Estimation of the optimal utilisation rates of radical prostatectomy, external beam radiotherapy and brachytherapy in the treatment of prostate cancer by a review of clinical practice guidelines.
]. Due to uncertainties in patient preference, sensitivity analysis considered a broad range of possibilities (e.g. 25–75% use of radical radiotherapy for early prostate cancer). This highlights the need for both better data on patient preference, patient access to decision tools, and where appropriate, radiotherapy consultation to ensure patient-centered decision making.
Though less influential overall, the greatest uncertainties for chemoradiation benefit related to uncertainties in the proportion of patients in the overall population who would be suitable candidates for concurrent chemotherapy. For example, there are limited data describing population-level performance status and concomitant comorbidities. Such information would benefit a wide range of population health services research, for example, allowing better adjustment of data investigating population-level treatment effectiveness using administrative data, and for better planning of cancer services based on evidence-based epidemiological models.
Conclusion
In conclusion, this report describes a robust, transparent and adaptable means for using publicly available data sources to estimate radiotherapy population benefits, and by extension, other forms of cancer treatment. Radiotherapy provides significant benefits to population-level cancer treatment programs, when used according to guidelines. The vast majority of radiotherapy benefit is derived from radiotherapy alone, with a modest additional benefit from concurrent chemotherapy.
Funding
There were no funding sources for this study.
Conflict of interest statement
The authors declare no conflicts of interest.
Acknowledgements
TP Hanna holds a research chair supported by the Ontario Institute for Cancer Research through funding provided by the Government of Ontario (#IA-035).
The population benefit of radiotherapy for cervical cancer: local control and survival estimates for optimally utilized radiotherapy and chemoradiation.
Barton MB, Jacob S, Shafiq J, Wong KH, Thompson S, Delaney G, et al. Review of optimal radiotherapy utilisation rates. http://tinyurl.com/pwkua34. Sydney: Ingham Institute, Collaboration for Cancer Outcomes Research and Evaluation; 2012.
National Health and Medical Research Council. NHMRC levels of evidence and grades for recommendations for developers of guidelines: Australian Government; 2009.
Therapy of locally unresectable pancreatic carcinoma: a randomized comparison of high dose (6000 rads) radiation alone, moderate dose radiation (4000 rads + 5-fluorouracil), and high dose radiation + 5-fluorouracil: The Gastrointestinal Tumor Study Group.
A randomized phase III study of radiotherapy alone or with 5-fluorouracil and mitomycin-C in patients with locally advanced adenocarcinoma of the pancreas: Eastern Cooperative Oncology Group study E8282.
Length and quality of survival after external-beam radiotherapy with concurrent continuous 5-fluorouracil infusion for locally unresectable pancreatic cancer.
Do V, Ng W, Delaney GP, Barton MB. An estimation of the population survival benefit of first-line chemotherapy for upper gastrointestinal cancer (#306) Clinical Oncology Society of Australia Annual Scientific Meeting; 2013.
Estimation of the optimal utilisation rates of radical prostatectomy, external beam radiotherapy and brachytherapy in the treatment of prostate cancer by a review of clinical practice guidelines.
32720-2&id=gr1.jpg){kind=link}
32720-2&id=gr2.jpg){kind=link}
32720-2&id=gr3.jpg){kind=link}