Chemotherapy remains the treatment of choice in symptomatic mCRPC, but survival benefits after undergoing chemotherapy are modest (on the order of a few months). In comparison to mitoxantrone (the prior standard chemotherapy agent), docetaxel was associated with better pain control, quality of life and more frequent PSA responses.  However, chemotherapy can also be associated with significant toxicity, with 18-44% rates of grade 3 or higher toxicity. National Cancer Institute Common Terminology Criteria for Adverse Events defines grade 3 as severe, grade 4 as life-threatening or disability and grade 5 as death.  Common toxicities from chemotherapy include neutropenia, generalized weakness, bone pain, fatigue, peripheral edema and mucositis. The most common grade 3 to 5 toxicities with docetaxel are: neutropenia, leucopenia, anemia, fatigue, infection and dehydration. 
Currently, there is a need to find tools that can help identify men who may be more or less likely to experience serious toxicity from chemotherapy because it could help during treatment decision-making. Predicting toxicities would help doctors determine the side effects and toxicities that specific patients might develop before prescribing the treatment. This way, it would make it easier for them to determine which treatment method would work, at which dose and method of delivery. Making a more informed decision can be important in this setting because of the increased risk of death or functional decline. It is especially helpful to be able to predict these toxicities in older adults because the risk of toxicity increases with age. In practice, chemotherapy is less likely to be given to older adults due to the concerns about their ability to tolerate it.  Many older adults tend to place an increasing value on avoiding treatments that adversely affect their quality of life or functional independence.  Since older adults have a higher risk of toxicity and place an increasing importance on quality of life, oncologists may find it harder to suggest the best treatment option. Hence, it would be useful to be able to predict toxicities from chemotherapy. This advancement in toxicity prediction would also help select up-front treatment modifications such as dose reduction or the addition of colony-stimulating factors to reduce toxicity.
Tools such as the Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) 5-point scale are currently used to determine risk by assessing a patient’s level of function and capability to perform self-care. Although this tool is a prognostic factor for survival and may help select which patients should not get chemotherapy, it is a poor predictor of toxicity risk because it is subjective, being subject to bias and high interobserver variability.  Oncologist judgement in stratifying patients into those at lower or higher risk of toxicity may be better, but it has rarely been formally compared against measures such as the ECOG PS. Finally, the agreement between currently used tools such as PS and clinical judgement by oncologists is still quite low. 
Our study sought to identify tools that could help inform treatment decision-making by improving the ability to predict a patient’s risk of chemotherapy toxicity. Distinguishing men at lower and higher risk of severe toxicity in men with mCRPC would help make better treatment decisions and allow a more informed decision about the risks and benefits of chemotherapy. In patients with very high risks of toxicity that may counterbalance any perceived benefits, there are four main options besides conventional dose chemotherapy: (a) reduced-dose chemotherapy; (b) use of colony-stimulating factors to reduce neutropenia and related complications; (c) alternative, gentler agents or clinical trials of novel therapies; (d) best supportive care. While our study did not focus on which treatment might be best, we sought to validate the Vulnerable Elders Survey-13 (VES-13) and Cancer and Aging Research Group (CARG) tool in mCRPC with the goal of helping a clinician’s judgment.
The VES-13 is a brief (3-4 minutes), self-report tool that measures vulnerability. The initial purpose of developing this tool was to better screen older persons at risk of health deterioration.  In the original study, vulnerable older people were defined as persons age 65 and older who were at increased risk of functional decline or death over 2 years.  The instrumental activities of daily living (IADLs) and activities of daily living (ADLs) that the VES-13 focuses on include shopping, performing light housework, managing finances, preparing meals, using the telephone, bathing, dressing, transferring, toileting, walking across the room, and eating.  However, its ability to predict grade 3-5 chemotherapy toxicity has yet to be studied.
The CARG tool uses a combination of 11 parameters, including age, tumor and treatment characteristics, laboratory data, and specific geriatric assessment parameters to help predict grade 3-5 chemotherapy toxicity in older patients with cancer. It categorizes people into low, intermediate and high risk of severe chemotherapy toxicity, in our case grade 3+ chemotherapy toxicity. It does include a geriatric assessment questionnaire with 6 domains: functional status, co-morbidity, psychological state, social activity, social support, and nutrition. The purpose of developing the CARG tool was to identify risk factors for chemotherapy toxicity in older adults undergoing various chemotherapy regimens and create a user-friendly risk stratification schema for chemotherapy toxicity.  The CARG tool was derived from a study of 500 patients undergoing a variety of chemotherapy regimens for various solid tumors. The CARG tool was recently validated externally  and helps to identify patients at greatest risk of chemotherapy toxicity. Although the CARG tool has been proven in a mixed cohort of patients with various cancers, there are no validation data for patients with mCRPC, and only 10% of the patients in the original study had genitourinary cancers.  Since different chemotherapy regimens have different toxicity risks, it is important to validate such tools in a more homogeneous cohort to ensure findings are similar to mixed cohorts.
For our study, we had each patient’s medical oncologist rate the patient’s risk of chemotherapy toxicity on a 10-point scale. “Oncologists are left with little guidance when it comes to identifying risk factors other than chronologic age or performance status, neither of which has been shown to predict well in heterogeneous older adult populations.” 
We recruited men aged 65 or older with mCRPC who were starting either first-line chemotherapy (receiving chemotherapy for the first time) or second-line chemotherapy (stopped first-line chemotherapy because of disease progression, toxicity, or other reasons). All but two (4%) participants received docetaxel-based chemotherapy, and the majority (n=29, 63%) received the standard dose of 75 mg/m2 every 3 weeks. Ten (22%) received a dose of 60 mg/m2, whereas 5 (11%) received a lower dose than this. Subjects were recruited either prior to starting chemotherapy or within the first two cycles as long as there was no dose reduction. Men unable to come for study visits or with a life expectancy of less than 3 months, a major neuropsychiatric abnormality, or limited English were excluded from the study.
We collected socio-demographic and medical information on all subjects at baseline, as well as physical performance measures (grip strength, timed up and go, and timed chair stands). The CARG and VES-13 tools were administered as well. The CARG toxicity prediction model was used to stratify patients into three groups (low, intermediate, and high risk) based on risk for grade 3+ chemotherapy toxicity. The VES-13 was used to measure vulnerability, which was defined by a score of 3 or greater. This cut-off point follows the conventional scoring system, but we also examined cut-offs of 2 or greater and 4 or greater. We also asked each subject’s treating physician to provide an estimate of the risk of chemotherapy toxicity on a scale from 1 (lowest risk) to 10 (highest risk). Oncologists were not told the results of the other assessment tools used in the study.
Following the baseline visit, follow-up assessments were performed after each cycle of chemotherapy (every 3 weeks) and after the final cycle. At each visit, a trained research coordinator recorded chemotherapy-related toxicities using the National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 (NCI CTCAE v4). Laboratory-based toxicities such as neutropenia were based on blood tests performed every three weeks. These same procedures were followed to record toxicity for men who were recruited after already having started chemotherapy, including for cycles administered before being enrolled on the study.
Sample sizes were based on the assumption that we would see the same rate of toxicity as in the original CARG study (i.e. 30% risk of grade 3+ toxicity for the low risk group, 52% for intermediate, and 83% for high)  and that equal proportions of patients would be enrolled in each risk group (i.e. one-third for each). Based on these assumptions, we calculated that we would require 45 patients.
46 men were recruited for the study with a mean age of 75. These participants had a median PSA level at baseline of 243.7 ng/mL and had relatively few other major medical problems (median Charlson Comorbidity Index score of 0). Although participants had a fairly high performance status (mean Karnofsky score of 77%), 50% were considered vulnerable based on the VES-13. Based on the CARG tool, only 2 (4%) patients were considered low risk, 29 (63%) were intermediate, and 15 (33%) were high risk of severe chemotherapy toxicity.
Grade 3+ and grade 2 chemotherapy toxicity were experienced by 20% and 67% of patients, respectively. The most common grade 3-5 toxicities were neutropenia (30%), generalized weakness (23%), and bone pain (15%), and the most common grade 2 toxicities were fatigue (35%), peripheral edema (7%), and mucositis (7%).
Grade 3+ toxicity was observed in 0 (0%), 5 (17%) and 4 (27%) patients in low, intermediate, and high CARG risk groups respectively, suggesting an incremental pattern across risk groups. However, this pattern was not statistically significant (p = 0.65). 22% of patients considered vulnerable by the VES-13 experienced grade 3+ toxicity, compared to 17% of patients considered non-vulnerable (p = 0.71). Age, comorbidity, Karnofsky performance score, and baseline physical performance measures did not seem to be predictors of grade 3+ toxicity. In addition, oncologist judgment of toxicity risk was a relatively poor predictor of actual toxicity.
The ability of the CARG tool to predict grade 2 toxicity appeared to be higher than the ability of the VES-13 to predict these toxicities, but this was not statistically significant, likely due to our small sample size (p = 0.072 for CARG, 0.75 for VES-13). Limiting the analyses to only those participants who were recruited prior to starting chemotherapy did not alter the findings.
The rates of grade 3+ toxicity found in our cohort were relatively low overall: only 20% compared to the 53% observed in the original CARG study. The same pattern was found in the three individual risk groups, with lower rates of toxicities observed in each compared to the original CARG study. However, the rate of toxicity in our cohort was similar to rates reported in other studies of older men with mCRPC. For example, the TAX327 trial by Tannock et al. reported severe adverse events in 26% of subjects, and grade 3+ neutropenia in 32%. 
Although we did not find statistically significant results for either of the tools tested, we did observe three key findings in our study. First, the risk of grade 3+ toxicity with docetaxel-based chemotherapy in the mCRPC population is lower overall and across CARG risk groups compared to the rates observed in the original study, which used data from patients with a variety of cancers. However, we still found that there was a gradient of toxicity risk across the different CARG risk groups (i.e. 0% in low, 17% in moderate, and 27% in the high risk group). Therefore, there is a need for further validation studies conducted with older men with mCRPC.
Second, our data on the performance of the VES-13 are the first in this population. Even though our findings were negative, we believe they warrant further investigation because of the ease of use and emerging value of the VES-13 in other geriatric oncology settings (e.g. 12). Third, we also provided the first data looking at oncologist judgment of toxicity risk, and compared that to the CARG and VES-13 tools. For tools to be useful in a busy clinical setting, they must provide better predictive ability than the usual clinical care. Therefore, further investigation in this area is important.
Some other limitations include the fact that we conducted our study at a single academic cancer center, limiting generalizability, and did not use the CRASH tool, another popular tool for predicting toxicities.  Future studies should directly compare the CRASH and CARG tools in the mCRPC setting. Lastly, the 10-point rating scale we used for oncologist predictions has not been validated in this context, and we did not provide any numerical anchors. Therefore, the different ratings may have meant different things to different oncologists. Further investigation is warranted in these areas.
In summary, toxicity with docetaxel in a cohort of older men in usual clinical practice was lower than predicted by the CARG tool. Although the CARG tool appeared to differentiate those at lower versus higher risk of chemotherapy toxicity and was better than clinician judgement or ECOG PS, a larger validation study is needed.
Written By: Thavalis Ja, Rathore Ma, Breunis Ha, Alibhai SMHa,b,c
a. Department of Medicine, University Health Network
b. Department of Medicine, University of Toronto
c. Institute of Health Policy, Management and Evaluation, University of Toronto
- American Cancer Society. Key statistics for prostate cancer [Internet]. American Cancer Society Inc.; 2016 [updated 2017 Jan 5]. Available from https://www.cancer.org/cancer/prostate-cancer/about/key-statistics.html
- Nussbaum N, George DJ, Abernethy AP, Dolan CM, Oestreicher N, Flanders S, Dorff TB. Patient experience in the treatment of metastatic castration-resistant prostate cancer: state of the science. Prostate Cancer and Prostatic Diseases. 2016 Jun 1;19(2):111-21.
- American Cancer Society. Prostate cancers [Internet]. American Cancer Society Inc.; 2016 [updated 2016 Mar 11]. Available from https://www.cancer.org/cancer/prostate-cancer/about/key-statistics.html
- Chi K, Hotte SJ, Joshua AM, North S, Wyatt AW, Collins LL, Saad F. Treatment of mCRPC in the AR-axis-targeted therapy-resistant state. Annals of Oncology. 2015 Oct 1; 26(10):2044-56.
- Tannock IF, de Wit R, Berry WR, Horti J, Pluzanska A, Chi KN, Oudard S, Théodore C, James ND, Turesson I, Rosenthal MA. Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. New England Journal of Medicine. 2004 Oct 7; 351(15):1502-12.
- Hurria A, Togawa K, Mohile SG, Owusu C, Klepin HD, Gross CP, Lichtman SM, Gajra A, Bhatia S, Katheria V, Klapper S. Predicting chemotherapy toxicity in older adults with cancer: a prospective multicenter study. Journal of Clinical Oncology. 2011 Aug 1; 29(25):3457-65.
- Rose JH, O'Toole EE, Dawson NV, Lawrence R, Gurley D, Thomas C, Hamel MB, Cohen HJ. Perspectives, preferences, care practices, and outcomes among older and middle-aged patients with late-stage cancer. Journal of Clinical Oncology. 2004 Dec 15; 22(24):4907-17.
- Kelly CM, Shahrokni A. Moving beyond Karnofsky and ECOG performance status assessments with new technologies. Journal of Oncology. 2016 Mar 15; 6186543.
- Sørensen JB, Klee M, Palshof T, Hansen HH. Performance status assessment in cancer patients. An inter-observer variability study. British Journal of Cancer. 1993 Apr; 67(4):773-5.
- Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, Roth C, MacLean CH, Shekelle PG, Sloss EM, Wenger NS. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. Journal of the American Geriatrics Society. 2001 Dec 1; 49(12):1691-9.
- Hurria A, Mohile S, Gajra A, Klepin H, Muss H, Chapman A, et al. Validation of a Prediction Tool for Chemotherapy Toxicity in Older Adults With Cancer. Journal of Clinical Oncology. 2016 Jul 10; 34(20:2366-71.
- Luciani A, Ascione G, Bertuzzi C, Marussi D, Codeca C, Di Maria G, et al. Detecting disabilities in older patients with cancer: comparison between comprehensive geriatric assessment and vulnerable elders survey-13. Journal of Clinical Oncology. 2010 Apr 20; 28(12):2046-50.
- Extermann M, Boler I, Reich RR, Lyman GH, Brown RH, DeFelice J, et al. Predicting the risk of chemotherapy toxicity in older patients: The Chemotherapy Risk Assessment Scale for High-Age Patients (CRASH) score. Cancer. 2011 Nov 9; 118(13):3377-86.