Urologic Oncology: Seminars and Original Investigations ] (2013) 1–8

1078-1439/$ – see fronthttp://dx.doi.org/10.1016/j

☆Presented in part as aOrlando, FL February 201

* Corresponding authorE-mail address: gsonpa1Equal credit.

Original article

Association of the Charlson comorbidity index and hypertension withsurvival in men with metastatic castration-resistant prostate cancer☆

Jatinder Goyal, M.D.a,1, Gregory R. Pond, Ph.D.b,1, Matthew D. Galsky, M.D.c,Ryan Hendricks, M.D.c, Alexander Small, M.D.c, Che-Kai Tsao, M.D.c, Guru Sonpavde, M.D.d,*

a Department of Medicine, The University of Alabama at Birmingham, Birmingham, ALb Department of Oncology, McMaster University, Hamilton ON

c Division of Hematology and Medical Oncology, Tisch Cancer Institute, The Mount Sinai School of Medicine, New York, NYd Department of Medicine, Section of Hematology and Oncology, The University of Alabama at Birmingham (UAB) Comprehensive Cancer Center,

Birmingham, AL

Received 24 December 2012; received in revised form 25 February 2013; accepted 28 February 2013

Abstract

Objectives: The independent prognostic effect of comorbidities on outcomes in men with metastatic castration-resistant prostate cancer(mCRPC) is unclear. We sought to determine whether the Charlson comorbidity index (CCI) and hypertension (HTN) are associated withoverall survival (OS) independent of known clinical prognostic factors in mCRPC.Patients and methods: A retrospective analysis was conducted on 221 patients with mCRPC treated with docetaxel plus prednisone

combined with AT-101 (bcl-2 antagonist) or placebo on a prospective randomized phase II trial. The Cox regression analysis was performedto identify whether the CCI or HTN or both (by medical history) independently predicted OS after adjusting for baseline variables known tobe associated with OS. The Wilcoxon rank sum test and the Fisher exact test were used to compare data by comorbidity groups (CCI as acontinuous variable, CCI ¼ 6 vs. CCI ≥ 7 and HTN vs. no HTN).Results: The CCI was 6 in 116 patients (52.7%), 7 in 70 (31.8%), 8 in 23 (10.5%), 9 in 4 (1.8%), and 10 in 7 patients (3.2%). HTN was

present in 107 (48.6%) patients. Patients with CCI of ≥7 were older and exhibited worse performance status and anemia than patients withCCI of 6 (P o 0.05). The CCI was not independently predictive of OS on univariable and multivariable analyses. HTN alone or incombination with the CCI was borderline significantly associated with OS (P �0.09) on both univariable and multivariable analyses.Conclusions: The CCI did not predict OS independent of known prognostic factors in mCRPC. Age, performance status, and anemia

may adequately capture comorbidities in the context of mCRPC, given their association with higher CCI. Further prospective study ofcomorbidities in a larger data set may be warranted. The study of HTN in a larger data set may also be warranted given its borderline-independent association with OS. r 2013 Elsevier Inc. All rights reserved.

Keywords: Charlson comorbidity index; Hypertension; Metastatic castration-resistant prostate cancer; Docetaxel; Overall survival; Progression-free survival

1. Introduction

Metastatic castration-resistant prostate cancer (mCRPC)is an aggressive disease with poor overall survival (OS) [1].Multiple baseline clinical and laboratory variables areassociated with survival, including age, performance status(PS), prostate-specific antigen (PSA), hemoglobin (Hb),

matter r 2013 Elsevier Inc. All rights reserved..urolonc.2013.02.015

poster at the Genitourinary Cancer Symposium,3.. Tel.: þ205-975-3742; fax: þ205-934-9511.vde@uabmc.edu (G. Sonpavde).

albumin, lactate dehydrogenase, alkaline phosphatase, Glea-son score, visceral metastases, pain, PSA doubling time,and bone scan and measurable disease progression [2,3].However, they only moderately account for the variabilityin survival seen in these patients.

The comorbidity burden is known to increase with ageand appears to adversely affect survival in localized prostatecancer [4–11]. Little is known, however, about the role ofcomorbidities in affecting outcomes in the more advanceddisease, especially mCRPC. If comorbidities confer anindependent effect on OS in mCRPC because of either anindependent adverse effect on survival or tolerability oftherapy or both, this has important implications when

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–82

designing and interpreting clinical trials, and for treatingpatients in general practice. The Charlson comorbidityindex (CCI) is one of the most widely used and convenienttools for the assessment of comorbidity burden [12].Notably, in the initial development of the CCI, hypertension(HTN) was one of the comorbidities evaluated, but wasfound to be not significantly associated with mortality inthat population and therefore was not included in the finalcomorbidity index. However, HTN is an important comor-bidity that is established to be associated with cardiovas-cular comorbidities and mortality. Moreover, androgensuppression is continued in patients with mCRPC andmay be associated with fatal myocardial infarctions [13].Thus, the administration of androgen deprivation in thosewith HTN may be postulated to increase cardiovascularmorbidity and mortality. Hence, we undertook this study todetermine whether CCI or HTN or both could independ-ently prognosticate outcomes in patients with mCRPC.

2. Materials and methods

2.1. Eligible patients

The medical history was recorded during study entryfor all 221 patients with mCRPC treated with docetaxeland prednisone plus either an investigational agent AT-101(bcl-2 antagonist) or placebo on a previously reportedrandomized phase II trial [14]. This trial was conductedwith approval from the Institutional Review Boards at therespective participating centers in the United States andRussia. One patient had rapid disease progression prior toreceiving any study treatment and was excluded from allfurther analyses. The trial accrued 116 patients in the UnitedStates and 104 patients in Russia and demonstrated nostatistical differences in OS, other secondary outcomes, andtoxicities between the 2 arms. The patients from both armsof the trial were therefore combined for analysis. In thistrial, patients were removed from therapy for progressionbased on Prostate Cancer Working Group (PCWG)-2guidelines and not for PSA progression alone [15]. Thestratification factors were pain and PS.

2.2. Measuring the CCI and HTN

Risk group stratification and PCWG-2 clinical subtypeinformation about all patients was also included foranalysis. We employed the comorbidity index developedby Charlson et al. to quantify baseline comorbiditiesavailable from the trial database [12]. This index is aweighted measure that incorporates 19 major diseasecategories and assigns a number to each condition accord-ing to its hazard of 1-year mortality. The weights are thensummed into a pooled score, which predicts survival for thepatient. According to the CCI, patients with a metastaticsolid tumor are assigned a score of 6; therefore, all patients

in this clinical trial have a minimum CCI score of 6.Although the optimal measure of comorbidities is unclear,the CCI is an established tool that has also been validated toassess comorbidity burden in patients with localized pros-tate cancer [16,17]. Given the nonsymmetric nature of theCCI, patients were categorized into 2 comorbidity groups(CCI = 6 vs. CCI ≥ 7). However, the CCI does not includeHTN, an important comorbidity, as a variable. Hence, wealso analyzed the presence or absence of HTN, as recordedin the patients' medical history, for association with OS.Patients also had blood pressure measured as part of theclinical trial screening process, and again on day 1 of cycle1, thus a second definition of HTN was given as systolicpressure ≥140 or diastolic pressure ≥90 as measuredon day 1.

2.3. Statistical analysis

The Wilcoxon rank sum test and the Fisher exact testwere used to compare data by comorbidity groups. Survivalwas calculated from the date of randomization to the date ofdeath. We used the Kaplan-Meier method to plot OS andprogression-free survival (PFS) of the different groups.Univariable and multivariable Cox proportional hazardsanalysis was performed to identify whether the CCI orHTN could prognosticate OS or PFS. Through this analysis,the CCI and HTN were adjusted for trial stratificationvariables (pain and PS), previously reported nomogram andrisk groups [2,3], and PCWG-2 clinical subtypes [15].Although treatment arm was not statistically different forany outcome, treatment arm was used as a stratificationfactor for all the Cox regression analyses. We did notperform statistical adjustment for multiple testing, as thiswas an exploratory analysis [18,19]. The nomogram con-tains factors for liver metastases, number of metastatic sites,pain, the Karnofsky PS, type of progression (measurabledisease progression or bone scan progression vs. PSAprogression), pretreatment PSA doubling time, PSA, tumorgrade, alkaline phosphatase, and Hb. The risk groupingcontains 4 factors: pain, visceral metastases, anemia, andbone scan progression. PCWG-2 clinical subtypes in thecontext of mCRPC consist of 3 groups: visceral metastasiswith or without other sites of disease, bone metastasis withor without soft tissue sites, and soft tissue or lymph nodemetastasis only [15]. The minimum possible CCI score was6 because of the presence of metastatic malignancy in allpatients. The CCI was evaluated as both a continuousvariable and a categorical variable (CCI¼ 6 vs. CCI ≥ 7).Statistical analysis was performed using the SAS andR software packages. Assuming a median OS of 18 monthsand median PFS of 10.5 months for patients with CCI ¼ 6or patients without HTN, the sample sizes in this study hadover 80% power to detect hazard ratios of 41.71 for OSand 1.57 for PFS. As a subanalysis, we also examined thedifferences based on the CCI and HTN in regard to number

Table 1Baseline characteristics of patients by comorbidity groups. Variables with significant differences (P o 0.05) are in bold

CCI ¼ 6 CCI ≥ 7 No hypertension Hypertension

n 116 104 113 107Overall population n (%) 220 54 (46.7) 56 (53.9) 60 (53.1) 50 (46.7)

Age Mean (SD) 220 68.2 (8.6) 70.5 (8.1) 67.2 (8.1) 71.5 (8.2)

Liver metastases n (%) 220 8 (6.9) 6 (5.8) 7 (6.2) 7 (6.5)

Visceral disease n (%) 220 33 (28.5) 23 (22.1) 28 (24.8) 28 (26.2)

Significant baseline pain n (%) 214 38/112 (33.9) 36/102 (35.3) 40/110 (36.4) 34/110 (32.7)

ECOG status 0 220 49 (42.2) 30 (28.9) 41 (36.3) 38 (35.5)1–2 67 (57.8) 74 (71.2) 72 (63.7) 69 (64.5)

Prior surgery n (%) 220 77 (66.4) 62 (59.6) 77 (68.1) 62 (57.9)

Prior radiotherapy n (%) 220 60 (51.7) 59 (56.7) 56 (49.6) 63 (58.9)

PSA progression n (%) 220 93 (80.2) 89 (85.6) 93 (82.3) 89 (83.2)

RECIST progression n (%) 220 26 (22.4) 26 (25.0) 26 (23.0) 26 (24.3)

Bone scan progression n (%) 220 54 (46.6) 41 (39.4) 52 (46.0) 43 (40.2)

Gleason score at diagnosis ≤7 220 48 (41.4) 35 (33.7) 43 (38.1) 40 (37.4)8–10 49 (42.2) 45 (43.3) 49 (43.4) 45 (42.1)Unknown 19 (16.4) 24 (23.1) 21 (18.6) 22 (20.6)

PSA, ng/ml Mean (SD) 218 389.6 (1,299) 343.8 (621) 462.1 (1,353) 268.5 (503)Median (range) 89.7 (0–13,056) 81.0 (0–3,631) 103.9 (0–13,056) 68.1 (0–2,593)

Alkaline phosphatise, μ/l Mean (SD) 220 200.2 (230) 291.8 (404) 238.3 (357.5) 249.0 (292.6)Median (range) 119.5 (44–1,483) 125.5 (45–2,933) 119 (44–2,933) 123 (45–1,653)

Baseline hemoglobin Mean (SD) 218 12.8 (1.7) 12.2 (1.7) 12.7 (1.7) 12.3 (1.7)Median (range) 12.9 (8.2–17.0) 12.3 (8.2–15.6) 12.8 (8.2–17.0) 12.3 (8.7–16.2)

Baseline albumin, g/l Mean (SD) 220 4.2 (0.7) 4.4 (0.7) 4.5 (0.7) 4.1 (0.6)Median (range) 4.1 (2.5–5.6) 4.6 (2.7–5.7) 4.6 (2.8–5.7) 4.1 (2.5–5.6)

PSA doubling time n (%) o55 d 220 58 (50.0) 64 (61.5) 63 (55.8) 59 (55.1)

Risk groups Poor risk 212 17/111 (15.3) 20/101 (19.8) 19/108 (17.6) 18/104 (17.3)Moderate risk 47 (42.3) 35 (34.7) 41 (38.0) 41 (39.4)Good risk 47 (42.3) 46 (45.5) 48 (44.4) 45 (43.3)

PCWG-2 clinical subtypes Visceral disease 220 33 (28.5) 23 (22.1) 28 (24.8) 28 (26.2)Bone � nodal disease 73 (62.9) 75 (72.1) 78 (69.0) 70 (65.4)Nodal disease only 10 (8.6) 6 (5.8) 7 (6.2) 9 (8.4)

ECOG ¼ Eastern Cooperative Oncology Group; RECIST ¼ Response Evaluation Criteria In Solid Tumors.

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–8 3

of cycles, all-grade adverse events (AEs), and AEs as areason for discontinuation.

3. Results

3.1. Patient characteristics

Patient demographics have been previously publishedand consisted of a typical population of patients withmCRPC [14]. A total of 220 patients were evaluable forour analysis (1 patient never received any therapy and wasconsidered inevaluable). A total of 116 patients (52.7%) had

a CCI score of 6, whereas 70 (31.8%), 23 (10.5%), 4(1.8%), and 7 (3.2%) had a score of 7, 8, 9, and 10,respectively (Table 1). HTN was identified in 107 patients(48.6%). Seventy-five patients (34.1%) had neither comor-bidity (CCI score 6) nor HTN. Patients with HTN had ahigher median CCI score than patients without HTN (7 vs.6.43, P o 0.001). Of 116 American patients, 45 (38.8%)had a CCI score ≥7 and 72 (62.1%) were identified ashaving HTN, compared with 59 of 104 (56.7%) Russianpatients with a CCI score ≥7 and 35 (33.7%) with HTN.Patients with CCI score ≥7 were significantly older, hadlower Hb, and had poorer PS, but also had higher serumalbumin level (P o 0.05).

Table 2Univariate and multivariate Cox regression analyses

Overall survival Progression-free survival

Type Hazards ratio (95% CI) P-value Hazards ratio (95% CI) P-value

Charlson comorbidity index Continuous 0.96 (0.78–1.18) 0.69 0.97 (0.81–1.16) 0.726 vs. ≥7 0.92 (0.63–1.35) 0.67 0.99 (0.72–1.37) 0.96

Charlson comorbidity index þ hypertension Continuous 1.03 (0.88–1.21) 0.68 1.03 (0.90–1.18) 0.676 vs. ≥7 1.44 (0.95–2.19) 0.087 1.22 (0.86–1.74) 0.27

Hypertension Present vs. absent 1.39 (0.95–2.04) 0.089 1.33 (0.96–1.85) 0.089

Charlson comorbidity index Continuous 0.93 (0.75–1.15) 0.49 0.93 (0.77–1.11) 0.41ECOG status stratum 1.30 (0.56–3.00) 0.54 1.48 (0.76–2.91) 0.25Baseline pain stratum 1.46 (0.97–2.20) 0.067 1.34 (0.95–1.90) 0.092

Charlson comorbidity index 0.89 (0.71–1.11) 0.29 0.95 (0.79–1.13) 0.55Nomogram 2.01 (1.53–2.63) o0.001 1.71 (1.35–2.16) o0.001

Charlson comorbidity index 0.95 (0.77–1.17) 0.63 0.99 (0.83–1.18) 0.93Risk groups 1.55 (1.21–1.99) o0.001 1.33 (1.06–1.67) 0.013

Charlson comorbidity index 0.97 (0.79–1.19) 0.75 0.98 (0.82–1.17) 0.83PCWG-2 category ¼ Bone � nodal disease 0.76 (0.49–1.17) 0.20 0.76 (0.52–1.09) 0.14PCWG-2 category ¼ Nodes only 0.72 (0.31–1.65) 0.43 0.87 (0.42–1.80) 0.71PCWG-2 category¼Visceral disease �other sites Reference

Charlson comorbidity index 6 vs. ≥7 0.88 (0.60–1.29) 0.51 0.92 (0.65–1.28) 0.61ECOG status stratum 1.31 (0.56–3.02) 0.53 1.47 (0.75–2.89) 0.27Baseline pain stratum 1.45 (0.97–2.18) 0.071 1.33 (0.94–1.87) 0.11

Charlson comorbidity index 0.84 (0.57–1.24) 0.37 1.00 (0.71–1.40) 0.99Nomogram 1.99 (1.52–2.60) o0.001 1.69 (1.35–2.13) o0.001

Charlson comorbidity index 0.96 (0.65–1.42) 0.84 1.03 (0.74–1.44) 0.85Risk groups 1.55 (1.21–1.99) o0.001 1.34 (1.07–1.68) 0.012

Charlson comorbidity index 0.96 (0.65–1.41) 0.82 1.03 (0.74–1.43) 0.88PCWG-2 category ¼ Bone � nodal disease 0.76 (0.49–1.18) 0.21 0.75 (0.52–1.09) 0.13PCWG-2 category ¼ Nodes Only 0.72 (0.31–1.67) 0.44 0.87 (0.42–1.80) 0.71PCWG-2 category ¼ Visceral disease � other sites Reference

Hypertension Present vs. 1.33 (0.91–1.96) 0.14 1.30 (0.93–1.83) 0.13ECOG status stratum absent 1.28 (0.56–2.94) 0.56 1.57 (0.80–3.08) 0.19Baseline pain stratum 1.38 (0.92–2.07) 0.12 1.23 (0.87–1.74) 0.24

Hypertension 1.42 (0.96–2.09) 0.076 1.42 (1.01–1.99) 0.042Nomogram 1.94 (1.50–2.53) o0.001 1.73 (1.38–2.16) o0.001

Hypertension 1.39 (0.95–2.04) 0.093 1.30 (0.93–1.82) 0.12Risk groups 1.55 (1.21–1.99) o0.001 1.34 (1.07–1.68) 0.012

Hypertension 1.38 (0.94–2.02) 0.097 1.33 (0.95–1.85) 0.093PCWG-2 category ¼ Bone � nodal disease 0.77 (0.50–1.18) 0.23 0.76 (0.53–1.09) 0.13PCWG-2 category ¼ Nodes only 0.70 (0.31–1.62) 0.41 0.87 (0.42–1.80) 0.71PCWG-2 category ¼ Visceral disease � other sites Reference

CI ¼ confidence interval; ECOG ¼ Eastern Cooperative Oncology Group.

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–84

3.2. Association of CCI score and outcomes

Table 2 shows the univariate and multivariate Coxregression analyses evaluating the role of CCI and HTNas a prognostic indicator for OS and PFS. The nomogramand risk grouping showed significant association with bothOS and PFS, whereas PCWG-2 subtypes did not. The CCI(as both a continuous and a categorical variables) was notfound to be a predictor of OS or PFS on univariate (Fig. 1Aand B) or multivariate analysis after adjusting for nomo-gram, risk grouping, or PCWG-2 subtypes. There were no

differences between those with CCI ¼ 6 and those withCCI ≥7 in terms of number of cycles, all-grade AEs, andAE as a reason for discontinuation (Table 3). However,those with CCI score ≥7 had less grade ≥3 AEs than thosewith CCI ¼ 6, although the median number of events wasonly 1 and 2, respectively.

3.3. Association of HTN and outcomes

HTN is of borderline significance on univariateanalysis for both OS (P ¼ 0.089) and PFS (P ¼ 0.089)

Fig. 1. Kaplan-Meier overall (A) and progression-free (B) survival curves of groups based on baseline CCI. (Color version of figure is available online.)

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–8 5

(Table 2; Fig. 2A and B). This borderline significance heldtrue after adjustment for nomogram (P ¼ 0.076 for OS andP ¼ 0.042 for PFS), risk grouping (P ¼ 0.093 and 0.12),and PCWG-2 subtype variables (P ¼ 0.097 and P ¼0.093) on multivariate analysis as well. When HTN isadded to CCI, the modified variable is of borderlinestatistical significance (P ¼ 0.087) as a prognostic factorfor OS on univariate analysis if used as a categoricalvariable (≥7 vs. 6). The result on multivariate analysiswas of borderline significance for OS but not for PFS (datanot shown). When examining HTN measured on day 1 ofcycle 1 (systolic ≥140 or diastolic ≥90), no association wasnoted on both univariable and multivariable analyses (datanot shown). The agreement of baseline blood pressuredefined HTN and with HTN defined by medical historyrecords was poor. Those with HTN received fewer cycles oftreatment (P ¼ 0.014), had more all-grade AEs (P ¼0.007), and had discontinued therapy more frequentlybecause of AEs (the Fisher exact test P-value ¼ 0.005)than those without HTN (Table 3). Those with HTN alsoexhibited more grade ≥3 AEs (P ¼ 0.009), although themedian number of events was only 2 and 1, respectively.

Table 3Adverse events and treatment duration based on CCI and HTN (Variables with

Number of cycles Median (interquartile range)Number of AEs Median (interquartile range)Number of grade 3 þ AEs Median (interquartile range)Number (%) reason for discontinuation Completed 17 cycles

AEAscenta requestDisease progressionInvestigator requestSubject requestOther

4. Discussion

This retrospective analysis of a large prospective phase IItrial data set of 220 men with mCRPC receiving docetaxel-based chemotherapy suggests that CCI score is not inde-pendently prognostic, whereas HTN appeared to haveborderline prognostic significance. The number of cycles,all-grade AEs, and AE as a reason for discontinuation wereall significantly higher in those with HTN (as opposed tothose without HTN), but not higher in those with higherCCI ≥7 (as opposed to those with CCI ¼ 6). Furtheranalysis of HTN in a larger data set may be warranted givenits borderline-independent association with OS, which maybe attributable to poorer tolerability of therapy. Age, PS,and anemia may adequately capture comorbidities in thecontext of mCRPC, given their association with higherCCI score. The caveat is that our findings were observed inthe context of a prospective trial and comorbidities maystill significantly affect outcomes in patients ineligiblefor a prospective trial for docetaxel-eligible patients.Moreover, certain comorbidities or therapeutic inter-ventions for comorbidities may negatively interact with

significant differences [P o 0.05] are in bold)

CCI ¼ 6 CCI ≥ 7 No hypertension Hypertension

9 (5–13) 8 (4–15) 10 (6–15) 7 (4–12)9 (3–16) 6 (2–13) 5 (2–12) 9 (4–16)2 (1–3) 1 (0–2.5) 1 (0–3) 2 (1–4)16 (13.8) 22 (21.2) 25 (22.1) 13 (12.1)27 (23.3) 21 (20.2) 16 (14.2) 32 (29.9)5 (4.3) 3 (2.9) 4 (3.5) 4 (3.7)26 (22.4) 35 (33.7) 33 (29.2) 28 (26.2)19 (16.4) 7 (6.7) 17 (15.0) 9 (8.4)20 (17.2) 13 (12.5) 17 (15.0) 16 (15.0)3 (2.6) 3 (2.9) 1 (0.9) 5 (4.7)

Fig. 2. Kaplan-Meier overall (A) and progression-free (B) survival based on baseline hypertension. (Color version of figure is available online.)

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–86

docetaxel-based chemotherapy and may continue to warrantattention.

There is only limited available literature on the prog-nostic effect of comorbidity burden on survival in prostatecancer. Indeed, comorbidity indices are not routinely usedto make treatment decisions or to stratify patients enrolledon clinical trials for any advanced malignancy. Comorbid-ities may confer competing causes for mortality (that mayinfluence the decision to intervene to treat malignancies) aswell as limit treatments because of poor tolerance fortoxicities. Multiple studies have demonstrated a relationshipbetween comorbidities and survival in patients with local-ized prostate cancer [4–11]. Generally, these studies dem-onstrate the association of higher comorbidities with higheroverall mortality but lower prostate cancer–specific mortal-ity in the context of earlier stages of disease [9,20,21].Conflicting data exist for the interaction of comorbiditiesand prostate cancer–specific mortality in the context ofscreening for prostate cancer [22,23]. Consistent data tosupport the effect of comorbidity status in more advanceddisease are also lacking. A study examining comorbiditiesin locally advanced T3a disease could not demonstrate anassociation of CCI with OS [24]. Moreover, in otheradvanced and aggressive malignancies such as pancreaticand colorectal cancer, the CCI was not consistently prog-nostic, again suggesting that comorbidities may not hold thesame relevance in the context of an advanced incurablecancer as long as PS allows therapeutic intervention[25,26].

Our understanding of the effect of comorbidities inpatients with mCRPC is incomplete. We sought to fill thisvoid by investigating whether comorbid disease burdenassessment using the CCI and HTN predicted survival. TheCCI is based on an end point of 1-year mortality, which isappropriate for our patient population of interest. We alsoincluded HTN in our analysis as the CCI does not include

this major comorbid condition in its weighted score. Ourresults show that regardless of the CCI score or history ofHTN, patients with mCRPC did not have any statisticallysignificant differences in overall or PFS. A large retro-spective study from a nationwide database reported that inthe subset of patients with distant metastases from prostatecancer, the hazards ratio for all-cause mortality was higherin patients with more comorbidities [5]. However, this studydid not clearly include patients with mCRPC. In anotherstudy, the authors reported that the baseline comorbiditynumber was an independent prognostic indicator of survivalin men with mCRPC receiving docetaxel-based chemo-therapy; the increase in hazard ratio for every unit ofincrease in comorbidity number was a modest 1.09 (95%confidence interval ¼ 1.04–1.14, P-value ¼ 0.0008) [27].

Some comments on the inherent strengths and limitationsof our study deserve discussion. Owing to the large andmulticentric nature of our trial enrolling a sizeable numberof patients and thorough follow-up, we were able to carryout a comprehensive prognostic analysis. However, ourresults should be interpreted keeping in mind the retro-spective nature of our study. The size of our studypopulation was limited and so is the statistical power ofour study. Also, the study population was a highly selectivegroup of patients who received treatment on a clinical trialand the results might not be applicable to the generalpopulation. The majority of patients (116 of 220, i.e.,52.7%) had no other major comorbidity other than theCCI points attributed to an advanced malignancy, that is,mCRPC. Patients with serious acute or uncontrolledcomorbidities are generally excluded from trials, even ifthe patients meet other eligibility criteria. Moreover, themechanism for the probable poorer tolerability of docetaxel-based therapy with HTN is unclear. More multicenterprospective trials with a larger number of patients areneeded in the future to effectively evaluate the effect of

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–8 7

comorbidities on survival. Finally, we investigated theeffect of the CCI score and HTN and could not evaluateother measures of comorbidities that may be potentiallybetter in patients with advanced malignancies. The AdultComorbidity Evaluation-27, the National Institute on Agingand National Cancer Institute, and American Society ofAnesthesiologists comorbidity indices have also beenemployed in other solid tumor settings, although it isunclear if a single optimal index exists [28,29]. All of theindices also include some comorbidities that have signifi-cantly diminished risk of mortality currently compared withthe time they were developed. Furthermore, other measuresof functioning may capture comorbidities better than thecomorbidity index at least in elderly patients, for example,the comprehensive geriatric assessment, activities of dailyliving score, and nutritional status [30].

5. Conclusions

The CCI score and HTN were not predictive for survival inpatients with mCRPC receiving docetaxel-based chemother-apy, although HTN demonstrated borderline significance andwas also significantly associated with a lower number ofcycles, higher all-grade AEs, and AEs as a reason fordiscontinuation. Age, PS, and anemia may adequately capturecomorbidities in the context of mCRPC, given their associa-tion with higher CCI. Prospective external validation in largedata sets of both comorbidities and HTN is warranted. At thistime, given either the lack of or modest association ofcomorbidities with OS in the context of mCRPC, patientswith controlled comorbidities and no docetaxel-specific contra-indications may continue to be offered and undergo docetaxel-based chemotherapy, provided the Eastern Cooperative Oncol-ogy Group PS is 0 to 2. In mCRPC, survival outcomes appearto be dominated by the course of the malignancy and not bycompeting risks from other comorbidities.

Acknowledgments

We would like to thank Ascenta Therapeutics forsupporting this study.

References

[1] Tannock IF, de Wit R, Berry WR, et al. Docetaxel plus prednisone ormitoxantrone plus prednisone for advanced prostate cancer. N Engl JMed 2004;351:1502–12.

[2] Armstrong AJ, Garrett-Mayer ES, Yang YC, de Wit R, Tannock IF,Eisenberger M. A contemporary prognostic nomogram for men withhormone-refractory metastatic prostate cancer: a TAX327 studyanalysis. Clin Cancer Res 2007;13:6396–403.

[3] Armstrong AJ, Tannock IF, de Wit R, George DJ, Eisenberger M,Halabi S. The development of risk groups in men with metastaticcastration-resistant prostate cancer based on risk factors for PSAdecline and survival. Eur J Cancer 2010;46:517–25.

[4] Lund L, Borre M, Jacobsen J, Sorensen HT, Norgaard M. Impact ofcomorbidity on survival of Danish prostate cancer patients, 1995–2006: a population-based cohort study. Urology 2008;72:1258–62.

[5] Berglund A, Garmo H, Tishelman C, Holmberg L, Stattin P, LambeM. Comorbidity, treatment and mortality: a population based cohortstudy of prostate cancer in PCBaSe Sweden. J Urol 2011;185:833–9.

[6] Post PN, Hansen BE, Kil PJ, Janssen-Heijnen ML, Coebergh JW.The independent prognostic value of comorbidity among menaged o 75 years with localized prostate cancer: a population-basedstudy. BJU Int 2001;87:821–6.

[7] Hall WH, Jani AB, Ryu JK, Narayan S, Vijayakumar S. The impactof age and comorbidity on survival outcomes and treatment patternsin prostate cancer. Prostate Cancer Prostatic Dis 2005;8:22–30.

[8] Guzzo TJ, Dluzniewski P, Orosco R, Platz EA, Partin AW, Han M.Prediction of mortality after radical prostatectomy by Charlsoncomorbidity index. Urology 2010;76:553–7.

[9] Albertsen PC, Moore DF, Shih W, Lin Y, Li H, Lu-Yao GL. Impactof comorbidity on survival among men with localized prostate cancer.J Clin Oncol 2011;29:1335–41.

[10] Nieder C, Dalhaug A, Pawinski A, Aandahl G, Norum J. Comorbid-ity, use of common medications, and risk of early death in patientswith localized or locally advanced prostate cancer. Sci World J2011;11:1178–86.

[11] Daskivich TJ, Chamie K, Kwan L, et al. Comorbidity and competingrisks for mortality in men with prostate cancer. Cancer2011;117:4642–50.

[12] Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method ofclassifying prognostic comorbidity in longitudinal studies: develop-ment and validation. J Chronic Dis 1987;40:373–83.

[13] D'Amico AV, Denham JW, Crook J, et al. Influence of androgensuppression therapy for prostate cancer on the frequency and timingof fatal myocardial infarctions. J Clin Oncol 2007;25:2420–5.

[14] Sonpavde G, Matveev V, Burke JM, et al. Randomized phase II trial ofdocetaxel plus prednisone in combination with placebo or AT-101, an oralsmall molecule Bcl-2 family antagonist, as first-line therapy for metastaticcastration-resistant prostate cancer. Ann Oncol 2012;23:1803–8.

[15] Scher HI, Halabi S, Tannock I, et al. Design and end points of clinicaltrials for patients with progressive prostate cancer and castrate levelsof testosterone: recommendations of the Prostate Cancer ClinicalTrials Working Group. J Clin Oncol 2008;26:1148–59.

[16] Kastner C, Armitage J, Kimble A, Rawal J, Carter PG, Venn S. TheCharlson comorbidity score: a superior comorbidity assessment toolfor the prostate cancer multidisciplinary meeting. Prostate CancerProstatic Dis 2006;9:270–4.

[17] Alibhai SM, Leach M, Tomlinson GA, Krahn MD, Fleshner NE,Naglie G. Is there an optimal comorbidity index for prostate cancer?Cancer 2008;112:1043–50.

[18] Poole C. Multiple comparisons? No problem! Epidemiology 1991;2:241–3.

[19] Rothman KJ. No adjustments are needed for multiple comparisons.Epidemiology 1990;1:43–6.

[20] Wo JY, Chen MH, Nguyen PL, et al. Evaluating the combined effectof comorbidity and prostate-specific antigen kinetics on the risk ofdeath in men after prostate-specific antigen recurrence. J Clin Oncol2009;27:6000–5.

[21] Briganti A, Spahn M, Joniau S, et al. Impact of age and comorbiditieson long-term survival of patients with high-risk prostate cancer treatedwith radical prostatectomy: a multi-institutional competing-risksanalysis. Eur Urol 2013;63(4):693–701.

[22] Crawford ED, Grubb R 3rd, Black A, et al. Comorbidity andmortality results from a randomized prostate cancer screening trial.J Clin Oncol 2011;29:355–61.

[23] Andriole GL, Crawford ED, Grubb RL 3rd, et al. Prostate cancerscreening in the randomized Prostate, Lung, Colorectal, and OvarianCancer Screening Trial: mortality results after 13 years of follow-up.J Natl Cancer Inst 2012;104:125–32.

J. Goyal et al. / Urologic Oncology: Seminars and Original Investigations ] (2013) 1–88

[24] Hsu CY, Joniau S, Oyen R, Roskams T, Van Poppel H. Correlationsbetween age, Charlson score and outcome in clinical unilateral T3aprostate cancer. Asian J Androl 2009;11:131–7.

[25] Vickers MM, Powell ED, Asmis TR, et al. Comorbidity, age andoverall survival in patients with advanced pancreatic cancer—resultsfrom NCIC CTG PA.3: a phase III trial of gemcitabine plus erlotinibor placebo. Eur J Cancer 2012;48:1434–42.

[26] Asmis TR, Powell E, Karapetis CS, et al. Comorbidity, age and overallsurvival in cetuximab-treated patients with advanced colorectal cancer(ACRC)—results from NCIC CTG CO.17: a phase III trial ofcetuximab versus best supportive care. Ann Oncol 2011;22:118–26.

[27] Halabi S, Kelly WK, George DJ, et al. Comorbidities predict overallsurvival (OS) in men with metastatic castrate-resistant prostate cancer(CRPC). J Clin Oncol 2011;29(Suppl. 7). [abstr 189].

[28] Hines RB, Chatla C, Bumpers HL, et al. Predictive capacity of threecomorbidity indices in estimating mortality after surgery for coloncancer. J Clin Oncol 2009;27:4339–45.

[29] Hurria A. Embracing the complexity of comorbidity. J Clin Oncol2011;29:4217–8.

[30] Caillet P, Canoui-Poitrine F, Vouriot J, et al. Comprehensive geriatricassessment in the decision-making process in elderly patients withcancer: ELCAPA study. J Clin Oncol 2011;29:3636–42.