Abstract: We analyzed adherence to multidisciplinary clinical pathways to understand the impact on the cost of care for patients with prostate cancer. Methods: A retrospective cohort design identified 560 patients with newly diagnosed prostate cancer between 2015 and 2017 who received all first-line treatment at Moffitt Cancer Center. The 1-year total cost of care by pathway adherence was examined by risk group and treatment received. Results: Pathway alignment was high (85%). Patients treated On-Pathway had lower risk disease and fewer comorbidities than those treated Off-Pathway (P<.01). One-year costs were significantly lower among patients treated On-Pathway ($8416) and for those in Low/Favorable-Intermediate, Unfavorable-Intermediate, and metastatic risk groups (P <.05). The distribution of treatment received differed by pathway alignment (P <.001). Costs differed significantly by treatment modality. Conclusion: Treatment selection (ie, surgery, radiation, active surveillance) was the main driver of total costs of care. Treatment considerations are based on multiple factors, including patient choice, risk group, and comorbidities. Clinical pathway-directed treatments, informed by medical evidence and patient choice, were associated with decreased costs of care.
Key Words: clinical pathways, prostate cancer, treatment costs, evidence-based care, health care expenditures, total cost of care
The cost of cancer care continues to rise, fueled by an ever-growing number of new, expensive drugs and other advances in technology thus reinforcing the need to use medical evidence as a basis for developing consensus-driven clinical care strategies.1-5 One major approach to maintain and improve the quality of oncology care is the incorporation of evidence-based care pathways into clinical practice. Ideally, pathways should support the complex and multimodality interactions innate to the management of cancer patients by emphasizing the interdisciplinary nature of treatment.6 Several recent studies have demonstrated that adherence to clinical pathways can result in improved clinical outcomes, quality indicators, and the cost-effectiveness of cancer care.7-13
In 2009, Moffitt Cancer Center (MCC) began its own clinical pathways program as part of a broader accountable care strategy. MCC’s pathways emphasize a multidisciplinary care model by incorporating treatment recommendations from surgeons, medical and hematologic oncologists, and radiation oncologists; guidance for laboratory, radiographic, and molecular diagnostic evaluations, monitoring and surveillance; clinical trials opportunities; and supportive or palliative care spanning the entire continuum of care. MCC also developed an electronic health record (EHR) integrated pathway tool which became available for use in patients with prostate cancer in 2015.”
The accountable care organization (ACO) analytics initiative at MCC was launched in 2013 to utilize data generated from clinical care processes to serve as the foundation for designing new alternative payment models within the center. We developed a platform to measure alignment to clinical pathways by integrating relevant clinical data from various sources within the cancer center to inform clinicians and health care administrators regarding performance. Our aim was to determine the effect of clinical pathways alignment on treatment-related costs. This article presents our retrospective study comparing the total cost of cancer care over 1 year for patients presenting to MCC between 2015 and 2017 with newly diagnosed adenocarcinoma of the prostate and who received all treatment at MCC. Evaluation includes patient and disease characteristics associated with pathways alignment and 1-year cost of care overall, by alignment status, risk group status, and treatment modality.
Materials and Methods
To assess the relationship between alignment with MCC’s clinical pathways and total cost of care, we conducted an institutional review board-approved retrospective cohort analysis of patients who were newly diagnosed with prostate cancer and who received their first course of therapy at MCC. Using Moffitt’s Cancer Registry data, we identified patients eligible for inclusion in the analysis if they had a diagnosis of adenocarcinoma of the prostate (International Classification of Diseases [ICD]-O-3 C61.9 with a histology code of 81403) between July 1, 2015 and June 30, 2017. Patients who were diagnosed with stage 0 disease, had a prior history of another reportable cancer, or did not receive all of their first course treatment at MCC were excluded. Based on this initial filter, 567 prostate cancer patients were identified and assessed for clinical pathway alignment (Figure 1).
Once the cohort of patients was identified, demographic, clinical, and billing information for patients was obtained from Moffitt’s Cancer Registry database, billing system, EHR, and clinical trials management system.
Determination of Clinical Pathway Alignment
The prostate cancer pathways were developed internally through a consensus-driven process that included
urologists, radiation oncologists, medical oncologists, and specialized clinical pharmacists. Treatment recommendations are based on medical evidence, national standards, and expert consensus. The pathways provide treatment recommendations for the majority of clinical scenarios, newly diagnosed patients by risk group, recurrence, and surveillance. Moffitt’s defined clinical pathways represent the most current treatment options based on prostate cancer risk group as summarized in Table 1.14 Classification of pathway alignment began by categorizing the eligible prostate cancer patients into nonmetastatic disease and metastatic disease risk groups according to the National Comprehensive Cancer Network (NCCN) categories for prostate cancer, using Cancer Registry Data.15 First-line therapy, as recorded in the Cancer Registry, was used to categorize patients by treatment modality received. Patients with nonmetastatic disease received treatment as follows: active surveillance, radiation therapy (RT Only), radiation therapy + hormonal therapy (RT +HT), surgery only, or surgery with adjuvant radiation therapy +/- hormonal therapy (Surgery +RT). The metastatic disease patients were categorized into two treatment modality groups: hormone therapy (Mets: HT) or all other, which consisted of chemotherapy, RT and/or surgery (Mets: Chemo/RT/Surg).
A rule-based algorithm was used to determine if the treatment a patient received was in alignment with MCC’s defined clinical pathway, taking into account treatment options, dosing, and specific timelines. For example, radiation pathways specify the total radiation dose by summing the regional dose and the boost dose, and exact dose delivered was required to be considered On-Pathway. Enrollment on a therapeutic clinical trial was also considered On-Pathway. Based on this algorithm, patients were categorized as On-Pathway (n=279), Off-Pathway (due to medical contraindication or patient choice as recorded in the Cancer Registry Data; n=7), or requiring further review by clinical staff to determine pathways alignment status and reasons for nonalignment (n=281; Figure 1).
Of the 281 patients whose charts were manually reviewed, seven patients were determined not to be eligible and excluded from the analysis (Figure 1). Of the remaining 274, 196 were classified as having received treatment in alignment with MCC pathways (On-Pathway) and 78 were classified as Off-Pathway. The clinical criteria that distinguish the risk groups used to define the MCC pathways (Table 1) were not always coded in the Cancer Registry with the granularity needed to accurately classify the patients into the Moffitt-defined risk groups. As such, the majority of patients who were determined to need further review were subsequently classified as On-Pathway following manual chart audit due to further refinement of their risk category (n=179). An additional 17 patients were classified as On-Pathway due to appropriate pathway recommended treatment requiring additional management for comorbidities or treatment-related toxicities (n=11) or enrollment on a clinical trial (n=6). Patients were classified as Off-Pathway because they received an alternative radiation treatment regimen (n=48) or hormone therapy regimen (n=11), had medical contraindications and/or an alternative patient choice (n=13), or had been classified in another risk group based on a different interpretation of the criteria (n=6). As a result, the final sample size was 560 men with prostate cancer.
Defining Treatment Costs
The hospital-based direct cost for each patient was determined using data from the hospital billing system. Patients who were diagnosed with prostate cancer at MCC presented to MCC a median of 14 days prior to their diagnosis date, and patients diagnosed at an outside facility presented to MCC 45 days following their diagnosis date. The timeframe for measuring cost was the 12-month period following the date the patient first presented for prostate cancer treatment at MCC. The date of presentation was selected as the starting timepoint for accrued costs over 1 year rather than the date of initial diagnosis to equalize variations in time-to-treatment. The cost analysis focused only on the MCC clinical services and activities provided to patients and did not include any professional charges.
Each risk group was evaluated separately for pathways alignment, and some risk groups with similar treatment options were combined to maximize sample size. The final nonmetastatic disease risk groups used in the analysis were: (1) Low/Favorable Intermediate (combined Very Low, Low, and Favorable Intermediate); (2) Unfavorable Intermediate; and (3) High (combined High and Very High) Risk Groups.
ICD-9-CM and ICD-10 billing codes were used to calculate the Charlson Comorbidity Index (CCI) for each patient.16 Demographic and clinical characteristics, including CCI, were described and compared between On-Pathway and Off-Pathway patients. The mean and SD were calculated for continuous variables, while frequencies and percentages were calculated for categorical variables. The Kruskal-Wallis, χ2 chi-square or Fisher’s exact test were used to compare patient characteristics, as appropriate, between the On-Pathway and Off-Pathway groups. Costs were compared between On- and Off-Pathway patients using generalized linear models (GLM) with a log link function, both unadjusted and adjusted for age, race, and CCI. All statistical analyses were conducted with the statistical significance level set to P<.05 and performed with SAS software version 9.3.17
A total of 560 patients diagnosed with prostate adenocarcinoma between July 1, 2015, and June 30, 2017, and who received all of their first course of therapy at MCC met the inclusion criteria and were included in this study. The patient characteristics for both On- and Off-Pathway groups are summarized in Table 2. No differences between pathway alignment were observed for race, ethnicity, or location of diagnosis. However, Off-Pathway patients presented with higher risk disease and had a significantly higher CCI compared to On-Pathway patients.
Overall, 85% of the patients were classified as On-Pathway, while 15% were Off-Pathway. The distribution of treatment modality received varied significantly for patients treated On-Pathway compared to those Off-Pathway (P<.001; Figure 2). Patients treated On-Pathway had a fairly even distribution of treatment by modality: surgery only (n=161, 34%), active surveillance (n=141, 30%), and RT +/- HT (n=130, 27%). Patients who were treated Off-Pathway primarily received RT +/-HT (n=71, 84%). After stratifying the pathway alignment analysis by risk group, On-Pathway patients with Low/Favorable Intermediate Risk received active surveillance (n=140, 51%), surgery only (n=65, 24%), or RT +/-HT (n=65, 24%), whereas all of the 36 Off-Pathway Low/Favorable Intermediate Risk patients received RT +/-HT. Among patients in the Unfavorable Intermediate and High-Risk Groups, On-Pathway patients were primarily treated with surgery only (n=96, 54%), and Off-Pathway patients mostly received RT +/-HT (n=35, 83%). Patients with metastatic disease treated On-Pathway mainly received hormone therapy only (n=14, 56%) and all patients treated Off-Pathway received more aggressive therapeutic regimens (n=8, 100%) (data not presented).
Average direct total costs of care for 1 year following initial presentation to MCC for patients with prostate cancer are presented for On- and Off-Pathway patients in
Table 3. Overall, the average direct cost for patients treated On-Pathway was $8416 lower than patients treated Off-Pathway (P<.0001), even after adjustment for age, race, and CCI. Significantly lower costs were observed for patients with metastatic disease treated On-Pathway compared with Off-Pathway ($22,473; P<.05). For patients with nonmetastatic disease, overall costs were significantly lower among patients treated On-Pathway compared with Off-Pathway ($7048; P<.0001), and pathways adherence was associated with lower costs for both the Low/Favorable Intermediate and Unfavorable Intermediate Risk Groups ($6849; P<.05 and $7277; P<.01, respectively.) However, no significant differences in cost were observed based on pathway alignment for the High-Risk Group.
Overall, 163 (29%) patients received surgery only and the utilization of surgery increased with advancing stage of disease: Low Risk 5%, Favorable Intermediate Risk 35%, Unfavorable Intermediate Risk 41%, and High Risk 48%. When compared to the average 1-year cost of care for patients receiving surgery only, active surveillance cost approximately 80% less, RT +HT cost 25% more, surgery +RT cost 80% more; these 1-year costs were significantly different (P<.001) (Figure 3). Among patients treated with metastatic cancer, those receiving Chemo/RT/Surg had almost three times higher costs compared to those that received hormone therapy only; these costs were significantly different from those of surgery only (P<.05).
The objective of this study was to evaluate the relationship between adherence to institutionally-developed, consensus-driven clinical pathways and the cost of care for patients with newly diagnosed prostate cancer. We found that overall adherence to clinical pathways was high (85%) and was associated with a significant cost advantage when the total direct cost of care at 1 year was compared with patients who were treated Off-Pathway. In the nonmetastatic group, the mean cost adjusted for age, race, and comorbidity was significantly lower for patients treated On-Pathway. Over 70% of the patients in this study had Low or Intermediate Risk features, and, in these groups, significantly lower costs were observed with pathway adherence. For patients with metastatic disease treated On-Pathway, the total cost of care was significantly lower, even when adjusted for age, race, and comorbidities. The main driver of cost was the choice of therapy. In general, patients treated in alignment with pathways recommendations received less costly treatment when compared to those patients who were treated Off-Pathway.
The adoption of clinical pathways provides an opportunity to instill greater accountability into the health care system by focusing on the use of medical evidence and treatment standardization as a foundation for high-quality clinical decisions. In the management of prostate cancer, especially in patients with newly diagnosed, early-stage disease, treatment options are based on multiple factors and involve a shared decision-making process that weighs the risks and benefits of therapy with an emphasis on patient choice.18,19 The presence of a robust clinical pathways program and concomitant development of clinical consensus regarding best practices should drive patient-centered care and ensure consideration of all therapeutic options including surgery, radiation, and active surveillance in newly diagnosed patients.20,21 Active surveillance is a reasonable option for many patients with prostate cancer given the natural history of the disease, since long-term follow-up of appropriately selected low- and intermediate-risk patients treated expectantly with active surveillance has been shown to result in comparable nonprostate cancer-specific mortality when compared to surgery or radiation therapy.22,23 Additionally, active surveillance with appropriate monitoring has been shown to be a cost-effective approach to the management of low-risk patients.24
In our study, pathways-driven recommendations for patients with nonmetastatic prostate cancer resulted in a fairly even distribution of therapeutic modality selection: radiation-based regimens (38%) vs surgery (35%) vs
active surveillance (27%). Active surveillance was particularly common in Low-Risk patients where 74% of these patients opted for this strategy. A relatively high proportion of Favorable Intermediate Risk group patients treated On-Pathway also were managed with active surveillance (28%), whereas all patients treated Off-Pathway in the Low/Favorable Intermediate Risk group received radiation-based regimens. In the Unfavorable Intermediate Risk group, the majority of patients who were considered On-Pathway received surgery only (57%), and Off-Pathway patients primarily received radiation-based regimens (89%: 59% RT only and 30% RT +HT). These differences in treatment modality selection resulted in decreased costs for patients with Low- or Intermediate-Risk disease.
As has been demonstrated by others,25 we saw that therapy choices also affected cost in patients presenting with metastatic cancer. More patients who were On-Pathway received hormone therapy only in comparison to patients treated Off-Pathway who were more likely to receive chemotherapy-containing regimens. Additionally, palliative therapy with radiation or surgery was more common in the Off-Pathway patients, accounting for higher costs.
The primary reason patients were found to be Off-Pathway was due to variation in radiation therapy regimens. The presence or absence of clinical comorbidities could be a confounding factor influencing treatment decisions. Overall, we saw that patients who were treated On-Pathway had a lower comorbidity index and stage of disease than the Off-Pathway group, suggesting that these patients may have required more individualized treatment considerations. Since the analysis was completed, radiation courses have evolved to include more hypofractionation.26
To be successful in decreasing cost, pathways should address all aspects of cancer management. This is the case with MCC’s multidisciplinary pathways, which offer recommendations along the entire continuum of care. Our study evaluated the total cost of care over 1 year rather than comparing the unit costs of each procedure in isolation to understand the overall influence of treatment decisions and their complications on cost. Others have demonstrated that, although drugs account for a significant percentage of total health care expenditures in the treatment of patients with cancer, other services such as inpatient care, surgery, radiation therapy, facility fees, and outpatient care are also key influencers, and the impact of these activities should be considered when understanding cost over time.2,4,27 Understanding cumulative costs and, especially, evaluating cost outliers could improve cost effectiveness by providing actionable information based on treatment patterns at the individual patient and physician level as well as the system level.
Mounting health care costs have been one of the primary motives driving the development of alternative payment models that incentivize quality outcomes and cost effectiveness. These models have identified barriers to implementation and have met with varying success.28-30 Adherence to evidence-based pathways recommendations, while resulting in improved clinical outcomes, also addresses appropriate utilization, which could result in cost savings or, at the very least, yield more predictable information regarding health care expenditures. Thus, pathways should be a core component of any accountable care strategy. Pathways have the potential to ensure behaviors that endorse appropriate treatment and avoid the risk of underutilization or overutilization of specific therapies or other resources. A recent study of urologists participating in an ACO offering shared savings based on financial performance demonstrated a lower likelihood of potential overtreatment in patients with newly diagnosed prostate cancer.31
Evidence-based clinical pathways define best practices and allow individualized treatment recommendations based on disease-specific and other relevant clinical factors. This study demonstrated that patients with prostate cancer treated in alignment with multidisciplinary clinical pathways had lower costs of care. In patients with both nonmetastatic and metastatic prostate cancer, costs were primarily associated with the treatment selected. Total cost of care evaluated over time provides a more comprehensive understanding of treatment-related costs, allowing institutions to evaluate clinical outcomes, cost efficiency, and quality. Clinical pathways encourage transparency and accountability by prospectively defining optimal care strategies and ensuring appropriate utilization of health care resources.
1. Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. Projections of the cost of cancer care in the United States: 2010-2020. J Natl Cancer Inst. 2011;103(2):117-128. doi:10.1093/jnci/djq495
2. Park J, Look KA. Health care expenditure burden of cancer care in the United States. Inquiry. 2019;56:1-9. doi:10.1177/0046958019880696
3. Howard DH, Chernew ME, Abdelgawad T, Smith GL, Sollano J, Grabowski DC. New anticancer drugs associated with large increases in costs and life expectancy. Health Aff (Millwood). 2016;35(9):1581-1587. doi:10.1377/hlthaff.2016.0286
4. Dieguez G, Ferro C, Pyenson BS. A multi-year look at the cost burden of cancer care. Milliman Research Report. Published April 11, 2017. Accessed January 7, 2021. https://www.milliman.com/en/insight/2017/a-multi-year-look-at-the-cost-burden-of-cancer-care
5. IQVIA INSTITUTE for Human Data Science. Global Oncology Trends 2018: Innovation, Expansion and Disruption. Published May 24, 2018. Accessed January 7, 2021. https://www.iqvia.com/insights/the-iqvia-institute/reports/global-oncology-trends-2018
6. Zon RT, Frame JN, Neuss MN, et al. American Society of Clinical Oncology Policy Statement on Clinical Pathways in Oncology. J Oncol Pract. 2016;12(3):261-266. doi:10.1200/JOP.2015.009134
7. Neubauer MA, Hoverman JR, Kolodziej M, et al. Cost effectiveness of evidence-based treatment guidelines for the treatment of non-small-cell lung cancer in the community setting. J Oncol Pract. 2010;6(1):12-18. doi:10.1200/JOP.091058
8. Rotter T, Kinsman L, James EL, et al. Clinical pathways: effects on professional practice, patient outcomes, length of stay and hospital costs. Cochrane Database Syst Rev. 2010;(3):CD006632. doi:10.1002/14651858.CD006632.pub2
9. Hoverman JR, Cartwright TH, Patt DA, et al. Pathways, outcomes, and costs in colon cancer: retrospective evaluations in two distinct databases. J Oncol Pract. 2011;7(suppl 3):52s-9s. doi:10.1200/JOP.2011.000318
10. van Hoeve J, de Munck L, Otter R, de Vries J, Siesling S. Quality improvement by implementing an integrated oncological care pathway for breast cancer patients. Breast. 2014;23(4):364-370. doi:10.1016/j.breast.2014.01.008
11. Feinberg BA, Milligan S, Cooper J, et al. Third-party validation of observed savings from an oncology pathways program. Am J Manag Care. 2013;19(SP4):153-157.
12. Hertler A, Chau S, Khetarpul R, et al. Utilization of clinical pathways can reduce drug spend within the oncology care model. JCO Oncol Pract. 2020;16(5):e456–e463. doi:10.1200/JOP.19.00753
13. Jackman DM, Zhang Y, Dalby C, et al. Cost and survival analysis before and after implementation of Dana-Farber Clinical Pathways for patients with Stage IV non-small-cell lung cancer. J Oncol Pract. 2017;13(4):e346-e352. doi:10.1200/JOP.2017.021741
14. Del Signore A. An in-depth look at Moffitt Cancer Center’s clinical pathways program and use in payer strategies—part I. J Clin Pathways. Published March 5, 2019. Accessed January 7, 2021. https://www.journalofclinicalpathways.com/depth-look-moffitt-cancer-centers-clinical-pathways-program-and-use-payer-strategies-part-i
15. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Prostate Cancer. Published August 19, 2019. Accessed January 7, 2021. https://www.nccn.org/professionals/physician_gls/pdf/prostate.pdf
16. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining Comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130-1139. doi:10.1097/01.mlr.0000182534.19832.83
17. SAS Institute Inc 2016. SAS/STAT 15.1, Cary, NC: SAS Institute Inc.
18. Bekelman JE, Rumble RB, Chen RC, et al. Clinically localized prostate cancer: ASCO clinical practice guideline endorsement of an American Urological Association/American Society for Radiation Oncology/Society of Urologic Oncology Guideline. J Clin Oncol. 2018;36(32):3251-3258. doi:10.1200/JCO.18.00606
19. Sanda MG, Cadeddu JA, Kirkby E, et al. Clinically localized prostate cancer: AUA/ASTRO/SUO Guideline. Part II: recommended approaches and details of specific care options. J Urol. 2018;199(4):990-997. doi:10.1016/j.juro.2018.01.002
20. Korman H, Lanni T Jr, Shah C, et al. Impact of a prostate multidisciplinary clinic program on patient treatment decisions and on adherence to NCCN guidelines: the William Beaumont Hospital experience. Am J Clin Oncol. 2013;36(2):121-125. doi:10.1097/COC.0b013e318243708f
21. Kuykendal AR, Hendrix LH, Salloum RG, Godley PA, Chen RC. Guideline-discordant androgen deprivation therapy in localized prostate cancer: patterns of use in the Medicare population and cost implications. Ann Oncol. 2013;24(5):1338-1343. doi:10.1093/annonc/mds618
22. Hamdy FC, Donovan JL, Lane JA, et al. 10-year outcomes after monitoring, surgery, or radiotherapy for localized prostate cancer. N Engl J Med. 2016;375(15):1415-1424. doi:10.1056/NEJMoa1606220
23. Klotz L, Vesprini D, Sethukavalan P, et al. Long-term follow-up of a large active surveillance cohort of patients with prostate cancer. J Clin Oncol. 2015;33(3):272-277. doi:10.1200/JCO.2014.55.1192
24. Sathianathen NJ, Konety BR, Alarid-Escudero F, Lawrentschuk N, Bolton DM, Kuntz KM. Cost-effectiveness analysis of active surveillance strategies for men with low-risk prostate cancer. Eur Urol. 2019;75(6):910-917. doi:10.1016/j.eururo.2018.10.055
25. Dragomir A, Dinea D, Vanhuyse M, Cury FL, Aprikian AG. Drug costs in the management of metastatic castration-resistant prostate cancer in Canada. BMC Health Serv Res. 2014;14:252. doi:10.1186/1472-6963-14-252
26. Konski A, Yu JB, Freedman G, Harrison LB, Johnstone PA. Radiation oncology practice: adjusting to a new reimbursement model. J Oncol Pract. 2016;12(5):e576-e583. doi:10.1200/JOP.2015.007385
27. Fitch KV, Pelizzari PM, Pyenson BS. Cost drivers of cancer care: A retrospective analysis of Medicare and commercially insured population claim data 2004-2014. Milliman Research Report. Published April 14, 2016. Accessed January 8, 2021. https://www.milliman.com/en/insight/cost-drivers-of-cancer-care-a-retrospective-analysis-of-medicare-and-commercially-insured
28. Kline RM, Brown M, Buescher N, et al. The Centers for Medicare & Medicaid Services oncology care model halfway through: perspectives from diverse participants. J Natl Cancer Inst. 2019;111(8):764-771. doi:10.1093/jnci/djz072
29. Walker B, Frytak J, Hayes J, Neubauer M, Robert N, Wilfong L. Evaluation of practice patterns among oncologists participating in the oncology care model. JAMA Netw Open. 2020;3(5):e205165. doi:10.1001/jamanetworkopen.2020.5165
30. Kline RM, Muldoon LD, Schumacher HK, et al. Design challenges of an episode-based payment model in oncology: the Centers for Medicare & Medicaid Services oncology care model. J Oncol Pract. 2017;13(7):e632-e645. doi:10.1200/JOP.2016.015834
31. Modi PK, Kaufman SR, Borza T, et al. Variation in prostate cancer treatment and spending among Medicare shared savings program accountable care organizations. Cancer. 2018;124(16):3364-3371. doi:10.1002/cncr.31573