- Email: [email protected]
Body composition and survival in the early clinical trials setting
European Journal of Cancer (2013) 49, 3068– 3075
Available at www.sciencedirect.com
journal homepage: www.ejcancer.com
Body composition and survival in the early clinical trials setting H. Veasey Rodrigues a,⇑, V.E. Baracos b, J.J. Wheler a, H.A. Parsons a, D.S. Hong a, A. Naing a, S. Fu a, G. Falchoock a, A.M. Tsimberidou a, S. Piha-Paul a, G. Chisholm c, R. Kurzrock a a
Department of Investigational Cancer Therapeutics (Phase I Clinical Trials Program), The University of Texas MD Anderson Cancer Center, Houston, TX, USA b Department of Oncology, University of Alberta, Edmonton, Canada c Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA
Available online 15 July 2013
KEYWORDS Body composition Sarcopenia Prognostic score Phase I clinical trial
Abstract Purpose: Delineate the relationships between body composition parameters, 90-day mortality and overall survival, and correlate them with known prognostic factors in an early clinical trials clinic. Patients and methods: We studied 306 consecutive patients with various tumours; body composition was analysed by computerised tomography images. Survival was measured from the first clinic visit, at 90-day period and until death/last follow-up visit. Results: Median patient age was 56 years; 159 patients were men. Ninety-day mortality rate was 12%. Median overall survival was 9 months. In multivariate analyses, high MD Anderson Cancer Center (MDACC) score (p < 0.0001) [lactate dehydrogenase (LDH) > normal, albumin < normal, Eastern Cooperative Oncology Group (ECOG) performance status > 1, metastatic sites > 2, gastrointestinal (GI) tumours], low skeletal muscle index (SMI) (p = 0.0406) and male gender (p = 0.0077) were independent predictors of poor survival. If Royal Marsden Hospital (RMH) score (LDH > normal, albumin < normal, metastatic sites > 2) was used in lieu of MDACC score, it was also significant (p = 0.0003). Including SMI and gender in the MDACC or RMH score improved the accuracy of the original model (p = 0.006 and p = 0.0037, respectively). Conclusion: Patients with low SMI have shorter survival. Gender and SMI strengthens the accuracy of MDACC or RMH scores as prognostic tools. Prospective validation of these findings is warranted. Ó 2013 Elsevier Ltd. All rights reserved.
⇑ Corresponding author: Address: Av. Europa, 105, Sa˜o Paulo, SP 01421-001, Brazil. Tel./fax: +55 11 30675400.
E-mail address: [email protected] (H. Veasey Rodrigues). 0959-8049/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejca.2013.06.026
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
3069
1. Introduction
2. Patients and methods
Early clinical trials are designed to determine metabolic and pharmacologic effects of drugs in humans, side-effects associated with increasing doses, maximum tolerated dose, and early evidence of efficacy and response. Patients referred to phase I clinical trials usually present with treatment-refractory, advanced malignancies, and many will not benefit from experimental treatments.1,2 Therefore, understanding prognostic factors can assist in determining possible outcomes. Personalised therapy that is, matching patients with agents based on an individualised approach is playing an increasingly important role in cancer therapy.3–6 The significance of host factors such as body composition to this field has not been well studied. Efforts have been made to delineate objective and reliable prognostic parameters. The Royal Marsden Hospital (RMH) has developed a prognostic scoring system using objectively measured parameters [>2 sites of metastasis, albumin below normal and lactate dehydrogenase (LDH) above normal] that have been validated in multi-institutional studies,7,8 including the phase I clinic at MD Anderson.9 Our group has also studied other prognostic factors related to survival in the phase I patient population. Among these are the gastrointestinal (GI) tumour type and poor Eastern Cooperative Oncology Group (ECOG) performance status and, when grouped with the RMH variables, the MD Anderson Cancer Center (MDACC) score10 may strengthen the ability of the RMH score to predict a poor outcome. Body weight and body composition (fat and lean mass) play a crucial role in the aetiology, prognosis and treatment outcomes of cancer.11,12 Sarcopenia, defined as absolute muscle mass below 2 or more standard deviations of the muscle mass in healthy young adults13 is closely associated with cancer cachexia, and has a significant impact on the prognosis of cancer and other chronic diseases.14–19 Patients referred to clinical trials usually present with progressive disease despite conventional management. Therefore, they are particularly susceptible to the end-organ effects of cachexia, reflected as changes in body weight and body composition. Clearly, in view of these factors, understanding how body weight and composition impact outcomes and survival in this population is important. In this study, we analysed body weight and body composition in 306 patients who were referred to the Clinical Center for Targeted Therapy (phase I clinic) of the MD Anderson Cancer Center. Our goal was to delineate the relationships between body composition parameters, including sarcopenia, 90-day mortality and overall survival, and correlate them with known prognostic factors.
A total of 306 consecutive patients with various advanced cancers referred to the phase I clinic starting in December 2004, who met protocol criteria for inclusion, were analysed. These patients were part of a larger pool of patients from a previous study that evaluated survival of patients in the phase I setting.10 Patients were included if the computed tomography (CT) imaging test had taken place within 5–50 days of their first visit to the phase I clinic. Images were used to assess body composition. All patients received treatment on early clinical trials including cytotoxic and/or targeted agents, systemic and/or local-regional therapy (intra-hepatic arterial infusion). The study was conducted according to guidelines of the Institutional Review Board at MD Anderson, and written informed consents for all investigational treatments were obtained. Patients were followed up to determine length of survival and relevant variables, including body composition. 2.1. Demographic and laboratory data Age, gender, cancer diagnosis, height and weight data were collected from patients’ electronic medical records (Table 1). Weight measurements were culled from the day closest to the date of CT imaging at the first visit to the phase I clinic; median time from weight to CT imaging was 1 day (range, 10 days before to 20 days after imaging). Laboratory values were collected from the closest day to the first visit to the clinic; median time from laboratory tests to first visit was 0 days (range, 15 days before to 15 days after the first visit). The median time between CT imaging and first clinic visit was 13 days (range, 0–50 days before visit), only 8.2% (25/ 306) of the patients had imaging done between 31 and 50 days prior to clinic visit. 2.2. Survival data Overall survival was measured from the day of the first visit to the phase I clinic until death or final follow-up visit. Patients still alive at last follow-up were censored at that date. Ninety-day survival was calculated by censoring alive patients at day 90. 2.3. Body composition assessments Body mass index (BMI) was calculated dividing patient’s weight (kg) by height (m2). Body composition (muscle and fat body area) was estimated using the validated method described below. Routine abdominal CT images at the level of the 3rd lumbar vertebra (L3) were chosen for analysis. The use
3070
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 Table 1 Patient characteristics. Women (147)
Men (159)
Total (306)
Age, years (range)
55 (16–76)
60 (24–84)
56 (16–84)
ECOG (%) 0 1 2 3 4 NAa
36 (24) 92 (63) 17 (12) 1 (0.5) 0 1 (0.5)
51 (32) 95 (60) 8 (5) 2 (1) 0 3 (2)
87 (29) 187 (61) 25 (8) 3 (1) 0 4 (1)
66.9 (28.9–150.5)
86.0 (52.5–161.3)
78.5 (28.9–161.3)
13 58 44 32
1 (1) 47 (29.5) 64 (40) 47 (29.5)
14 (5) 105 (34) 108 (35) 79 (26)
Body composition parameters, cm2/m2 (95% CI) Skeletal muscle index 40.9 (40.3–42.6) Adipose index 64.4 (63.7–83.4)
50.5 (49.4–52.6) 91.8 (83.8–103.6)
44.9 (45.7–47.3) 79.7 (77.0–91.2)
Sarcopenia (%) NAb
51 (35) 1
93 (59) 1
144 (47)
Diagnosis (%) Gastrointestinal Genitourinary Breast Thyroid Melanoma Head and neck Lung Gynaecological Sarcoma Othersc
47 (32) 2 (1) 30 (20) 16 (11) 7 (5) 5 (3) 10 (7) 14 (10) 5 (3) 11 (8)
52 (33) 33 (21) 1 (1) 14 (9) 16 (10) 15 (9) 8 (5) – 7 (4) 13 (8)
99 35 31 30 23 20 18 14 12 24
RMH prognostic score 0 1 2 3 NAa
48 (33) 64 (44) 27 (18) 7 (5) 1
71 (45) 56 (35) 28 (18) 4 (2) 0
119 (39) 120 (39) 55 (18) 11 (4) 1
MDACC prognostic score 0 1 2 3 4 5 NAa
8 (5) 47 (32) 44 (30) 30 (21) 14 (10) 3 (2) 1
23 (14) 47 (30) 45 (28) 31 (20) 11 (7) 2 (1) 0
31 (10) 94 (31) 89 (29) 61 (20) 25 (8) 5 (2) 1
No of trials 1–2 >2
139 (95) 8 (5)
143 (90) 16 (10)
282 (92) 24 (8)
Weight, kg 2
BMI, kg/m (%) <18.5 18.5–24.9 25–29.9 P30
(9) (39) (30) (22)
(32) (11) (10) (9) (8) (7) (6) (5) (4) (8)
BMI: body mass index; <18.5 kg/m2 = underweight, 18.5–24.9 kg/m2 = normal, 25–29.9 kg/ m2 = overweight, P30 kg/m2 = obese. ECOG: Eastern Cooperative Oncology Group performance status score. MDACC: MD Anderson Cancer Center. NA: not available. RMH: Royal Marsden Hospital. a ECOG performance status not documented at the first visit to phase I clinic. b Not available due to technical limitation on CT imaging. c Others (non-melanoma skin cancer, thymus, unknown primary, peritoneal, CNS, lymphoma).
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
of L3 as the landmark for body composition analysis has previously been described and validated against dual X-ray absorptiometry in healthy populations and in patients with advanced cancer.20–22 Muscle, intramuscular fat, subcutaneous fat and visceral fat were identified by a single assessor trained in the specific anatomy of these tissues, demarcated using previously described Hounsfield unit thresholds23,24 and quantified in crosssectional areas (cm2) by the SliceOMatic software, version 4.3 (Tomovision, Montreal, QC, Canada). Fat tissue analysis was done in its totality and also dividing into subcutaneous fat and visceral fat because there is evidence that visceral fat may be associated with worse outcome.25–27 Sarcopenia was defined by a lumbar skeletal muscle index (skeletal muscle area at L3 divided by the height squared) lower than 38.5 cm2/m2 for women and lower than 52.4 cm2/m2 for men, as previously described.28 Adipose index was determined by dividing adipose tissue area at L3 by the height squared. 2.4. End-points and statistical analyses The goal of this study was to investigate the impact of body composition parameters, including sarcopenia, as prognostic factors for 90-day mortality and overall survival in patients referred to early clinical trials, and compare results with prognostic scores. Descriptive statistics were used to summarise patients’ characteristics at baseline by gender. Differences in categorical variables were determined by chi-square and Fisher’s exact tests, when applicable. Differences in continuous variables were determined using a t-test. Cox regression was performed for 90-day and overall survival. Each variable was first tested at a 10% significance level in a univariate model. The significant variables were then tested in a ‘univariate’ model adjusted for age and gender. Variables in the adjusted univariate model, which were significant at the 10% level, were then tested together in multivariate analysis. Harrel’s C statistic29 was calculated for multivariate models and z-tests were performed to test relative predictive accuracy. Analyses were performed using SAS 9.2 (SAS Institute, Cary NC) and R (The R Project for Statistical Computing, Vienna, Austria). 3. Results 3.1. Patient characteristics Fifty-two percent of patients were men (159/306); median age was 56 years; and ECOG performance status was 0 or 1 for 90% of patients (274/305) (one patient did not have ECOG status recorded). Sixty-one percent (187/306) of patients were overweight or obese, 34% (105/306) were normal weight and 5% (14/306) were
3071
underweight. Sarcopenia was found in 47% (144/304) of patients (two patients had technical issues surrounding CT imaging and sarcopenia could not be evaluated, though fat area was evaluated). The most frequent diagnoses were GI (32%, 99/306), genitourinary (11%, 35/ 306) and breast cancers (10%, 31/306). Among GI tumours, colorectal represented the vast majority (69%, 68/99), followed by upper GI (13%, 13/99), pancreatic (11%, 11/99) and small intestine tumours (7%, 7/99). The median number of prior therapies was four (range, 0–12). Only less than half percent (13/306) were previously exposed to tyrosine kinase inhibitors such as sorafenib. Patients with good prognosis using RMH score (0–1) were 78% (239/306) and 70% (214/306) had a low-to-intermediate (0–2) MDACC risk score10 (Table 1). 3.2. Clinical trials Patients were treated in 77 clinical trials, 92% (282/ 306) of patients participated in two or fewer trials (Table 1). Most of the trials consisted of targeted drugs, either as single agents (56%, 43/77) or in combination (13%, 10/77). The remainder trials consisted of combinations of cytotoxic and targeted drugs (15%, 12/77), single cytotoxic agents (12%, 9/77) and combinations of cytotoxic agents (4%, 3/77). 3.3. 90-Day mortality The 90-day mortality rate was 12% (36/306). The following categorical variables were evaluated as predictors of 90-day mortality: gender (female versus male), RMH score (P2 versus <2), albumin (<3.5 g/dL versus P3.5 g/dL), LDH (>618 IU/L versus 6618 IU/L), GI tumours (yes versus no), ECOG status (P1 versus 0), number of metastatic sites (P3 versus <3) and sarcopenia (yes versus no). The following continuous variables were evaluated as predictors of 90-day mortality: age, BMI, weight, MDACC score, skeletal muscle area (SMA), skeletal muscle index (SMI), visceral adipose area (VAA), subcutaneous adipose area (SCAA), total adipose area (TAA), adipose index (AI). 3.3.1. Univariate analysis Albumin <3.5 g/dL (p = 0.0163), low BMI (p = 0.0002), GI tumour (p = 0.0442), LDH > 618 IU/L (p = 0.0139), high MDACC score (p = 0.0001), metastatic sites P3 (p = 0.0136), RMH score P2 (p = 0.0022), low SCAA (p = 0.0146), low TAA (p = 0.0431) and low weight (p = 0.0014) were associated with a higher 90-day mortality (Table 2). These values were adjusted to age and gender, and adjusted TAA was not significant (p = 0.0793), whereas all variables mentioned above remained significant predictors for 90-day mortality. Because low SMI showed a trend
3072
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
Table 2 Univariate and multivariate analysis of predictor variables of 90-day mortality. Variables
HR
p Value
Univariate model Age* AI* (cm2/m2) Albumin (<3.5 g/dL versus >3.5 g/dL) BMI* (kg/m2) ECOG status (P1 versus <1) Gender (female versus male) GI tumour (Y versus N) LDH (>618 IU/L versus <618 IU/L) MDACC score* Metastatic sites (P3 versus <3) RMH score (P2 versus <2) Sarcopenia (Y versus N) SCAA* (cm2) SMA* (cm2) SMI* (cm2/m2) TAA* (cm2) VAA* (cm2) Weight* (kg)
0.982 0.997 2.521 0.874 1.750 1.121 1.958 2.273 1.729 2.302 2.819 1.678 0.995 0.993 0.966 0.998 0.997 0.969
0.1523 0.3276 0.0163 0.0002 0.1839 0.7322 0.0442 0.0139 <0.0001 0.0136 0.0022 0.1338 0.0146 0.1511 0.0793 0.0431 0.1147 0.0014
Multivariate model – Composite MDACC score 0.976 Age* BMI* 0.939 Gender (female versus male) 0.734 1.794 MDACC score* SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.996
0.0816 0.5194 0.4679 0.0002 0.7455 0.8921
Multivariable model – MDACC score individual variables 0.976 Age* Albumin (<3.5 g/dL versus > 3.5 g/dL) 1.913 BMI* (kg/m2) 0.929 Gender (female versus male) 0.731 GI tumour (Y versus N) 2.055 LDH (>618 IU/L versus <618 IU/L) 2.230 Metastatic sites (P3 versus <3) 1.612 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.992
0.0840 0.1371 0.4552 0.4656 0.0530 0.0222 0.2023 0.8034 0.8078
AI, adipose index; BMI, Body mass index; ECOG, Eastern Cooperative Oncology Group performance status score; GI, gastrointestinal; LDH, lactate dehydrogenase; MDACC, MD Anderson Cancer Center; SCAA, subcutaneous adipose area; SMA, skeletal muscle area; SMI, skeletal muscle index; TAA, total adipose area; VAA, visceral adipose area. * Continuous variables.
(p = 0.0793) to higher 90-day mortality, it was included in the adjusted univariate model and was found to be a significant predictor (p = 0.0338). 3.3.2. Multivariate analysis The multivariable analysis was performed with the significant variables from adjusted univariate model avoiding component variables (e.g. weight is a component of BMI). A high MDACC score was the only variable associated with increased 90-day mortality (p = 0.0002). If the analysis was done using the individual components of the MDACC score, only LDH > 618 IU/L (p = 0.0222) was significant. GI tumours showed a trend towards significance as a predictor of 90-day mortality (p = 0.0530) (Table 2).
If RMH score was used instead of MDACC score, it was also selected as predictor of 90-day mortality (p = 0.0126) along with GI tumour (p = 0.0271). 3.4. Overall survival The median overall survival was 9.0 months (IC 95% 7.2–10.8). The same variables evaluated as predictors of 90-day mortality were evaluated for overall survival. 3.4.1. Univariate analysis The univariate model showed that albumin <3.5 g/dL (p = 0.0092), ECOG P1 (p = 0.030), GI tumour (p = 0.0166), LDH >618 IU/L (p < 0.0001), high MDACC score (p < 0.0001), metastatic sites P3 (p = 0.001), RMH score P2 (p < 0.0001), sarcopenia (p = 0.0138), low SCAA (p = 0.0362) and low SMI (p = 0.0359) were associated with a shorter survival. These variables all remained significant after adjustment for age and gender. We have also analysed the overall survival according to body mass index and low skeletal muscle (sarcopenia). We found that there is a trend (p = 0.06) to a better survival for no sarcopenic patients, especially if BMI > 25 kg/m2 (11.5 months, 95% CI 9.1– 13.9), if BMI < 25 kg/m2 median survival was 8.6 months (95% CI 7.1–10.1). Sarcopenic patients had a median survival of 7.8 (95% CI 5.0–10.6) and 7.0 months (95% CI, 5.2–8.7) if BMI > 25 kg/m2 and BMI 6 25 kg/m2, respectively. 3.4.2. Multivariate analysis The multivariate analysis was performed in different models to avoid having component variables (e.g. the individual variables of the MDACC score) analysed together. High MDACC score (p < 0.0001), low SMI (p = 0.0406) and male gender (p = 0.0077) were independently associated with shorter survival (Table 3). When the MDACC score components were analysed individually, male gender (p = 0.0071), LDH > 618 IU/ L (p < 0.0001) and low SMI (p = 0.0445) were associated with shorter survival. ECOG performance status P1 showed a trend to shorter survival (p = 0.0617) (Table 3). Sarcopenia as a dichotomous variable (SMI < 52.4cm2/m2 for men and <38.5cm2/m2 for women) was not selected as an independent variable. The RMH score was not included in the multivariate analysis because it is included in the MDACC score, but RMH score (p = 0.0003) is also selected as predictor of overall survival if used instead of the MDACC score in the multivariate model, along with low SMI (p = 0.0300) and male gender (p = 0.0093). 3.4.3. Including SMI in composite models to predict survival SMI was included in the composite scores to investigate whether this variable increased the accuracy of the
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 Table 3 Univariate and multivariate analysis of predictor variables of overall survival. Variables
HR
p value
Univariate model Age* AI* (cm2/m2) Albumin (<3.5 g/dL versus >3.5 g/dL) BMI* (kg/m2) ECOG status (P1 versus <1) Gender (female versus male) GI tumour (Y versus N) LDH (>618 IU/L versus <618 IU/L) MDACC score* Metastatic sites (P3 versus <3) RMH score (P2 versus <2) Sarcopenia (Y versus N) SCAA* (cm2) SMA* (cm2) SMI* (cm2/m2) TAA* (cm2) VAA* (cm2) Weight* (kg)
1.000 0.999 1.579 0.983 1.332 0.861 1.357 2.016 1.363 1.491 2.006 1.340 0.999 0.998 0.987 0.999 0.999 0.997
0.963 0.264 0.0092 0.101 0.030 0.203 0.0166 <0.0001 <0.0001 0.001 <0.0001 0.0138 0.036 0.194 0.036 0.159 0.296 0.340
Multivariable model – Composite MDACC score 0.993 Age* Gender (female versus male) 0.634 MDACC score* 1.390 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.982
0.1525 0.0077 <0.0001 0.4365 0.0406
Multivariable model – MDACC score individual variables 0.992 Age* Albumin (<3.5 versus P3.5) 1.128 ECOG (>1 versus 0) 1.319 Gender (female versus male) 0.626 GI tumour (Y versus N) 1.218 LDH (>618 versus 6618) 2.142 Metastatic Sites (P3 versus <3) 1.238 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.981
0.1285 0.5572 0.0617 0.0071 0.1838 <0.0001 0.1518 0.1151 0.0445
AI, adipose index; BMI, Body mass index; ECOG, Eastern Cooperative Oncology Group performance status score; GI, gastrointestinal; LDH, lactate dehydrogenase; MDACC, MD Anderson Cancer Center; SCAA, subcutaneous adipose area; SMA, skeletal muscle area; SMI, skeletal muscle index; TAA, total adipose area; VAA, visceral adipose area. * Continuous variable.
composite model. A new model including MDACC score, gender and SMI has a Harrel’s C statistic of 0.62 and the MDACC score (only) model has a Harrel’s C statistic of 0.59. Harrel’s C statistic measures the general predictive power of a survival model.29 Thus, the new model has a higher Harrel’s C statistic and the difference is statistically significant at p = 0.006, indicating that the new model improves the predictive accuracy of the original MDACC score (only) model. The same is true for RMH score; the addition of SMI and gender improved the accuracy of the model compared to the RMH score (only) model (0.62 versus 0.56, p = 0.0037). SMI, however, did not improve the score when gender was excluded from the model.
3073
4. Discussion Patient selection in early anticancer drug development is important for several reasons, including to avoid harm and medical futility because of uncertainty regarding drug toxicity and activity, and avoid confounding factors that may negatively impact the drug safety (adverse event) profile and efficacy data. Several groups, including ours, have described composite scores based on objective parameters that can be used to guide therapy decision-making.7,8,10,30 Survival prognostic scores are also helpful for identifying patients at risk for early death (90-day mortality). Although patients referred to early clinical trials have advanced cancers, a large proportion of enrolled patients had favourable RMH prognostic scores (56– 79%).7,9 Seventy-eight percent of our patients had good RMH prognostic scores and 70% had low-to-intermediate MDACC risk scores.10 This result might be explained by referral bias, because it is known that clinical trials generally exclude severely ill patients, and/or the increased availability of clinical trials in tertiary cancer centres. Another aspect of a referral bias can explain the unusual prevalence of cancer subtypes in our population. Lung cancer was less frequent than thyroid cancer and melanoma for example, and may have influenced our findings, since some cancers are more related to cachexia than others. Sarcopenia was found in 47% of patients in concordance with our previous finding of 51% of sarcopenia in patients seen at MDACC-Phase I clinic.15 Other authors reported sarcopenia in 51% of patients with pancreatic cancer31 and 52.5% in patients with renal cancer.7 It has been well documented that exposure to sorafenib, a tyrosine kinase inhibitor, can promote muscle loss/sarcopenia,32 however, few of our patients (13/ 306) were previously treated with sorafenib or other VEGFR tyrosine kinase inhibitors. The majority (61%) our population was overweight or obese, a percentage consistent with other reports7,15; this is a frequent finding in advanced cancer patients,33,34 particularly those from developed nations where overweight and obesity are epidemic.35 The 90-day mortality in this study was evaluated from the day of the first visit to our clinic, and not the first day of the clinical trial on which a patient was enrolled, as reported by other groups.8,30 This could account for our slightly lower 90-day mortality, which was 12% in the current study versus the 15–20% reported previously.7,8,30 A high MDACC score10 was associated with increased 90-day mortality in multivariate analysis (Table 2). Body weight, BMI and other body composition parameters were not associated with increased 90-day mortality in our cohort when MDACC as a composite score was included. The individual component analysis showed that only LDH above normal
3074
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
was independently associated with increased 90-day mortality. If MDACC was substituted for RMH score it was also selected as an independent variable predicting 90-day survival, along with GI tumour type. Until now, the known prognostic scores (MDACC and RMH) have not considered body weight and body composition parameters as prognostic variables, although scientific evidence links cachexia–sarcopenia and sarcopenic obesity with cancer prognosis and survival.15,28,31,36–38 Interestingly, Kohara et al.39 demonstrated a relation between sarcopenic obesity and high levels of leptin, which is a pro-angiogenic factor, and could explain in part the inferior outcome of this group. This study demonstrated that low skeletal muscle index (SMI) as a continuous variable, was associated with shorter overall survival in a multivariate analysis model (Table 3), and it was an independent predictor of overall survival even in the presence of validated prognostic scores (MDACC and RMH) and other individual factors, such as LDH and ECOG performance status. Sarcopenia, which is the SMI variable dichotomised, was, however, not an independent factor predicting shorter survival in a multivariable model. The latter may be because the parameters defining the cutoff for sarcopenia are derived from an obese population,28 and did not include the full spectrum of cancer patients. The literature demonstrates that low skeletal muscle mass/sarcopenia is frequently associated with poor outcomes. For instance, in patients with colorectal cancer undergoing resection of liver metastasis, sarcopenia was associated with greater postoperative complications40 and worse progression-free survival and overall survival.41 In overweight or obese cancer patients, sarcopenia is an independent predictor of survival,15,28,31 and in patients with stage III melanoma sarcopenia was an independent factor for disease-free survival.42 Finally, our study demonstrates that addition of SMI and gender in a model with MDACC score10 or RMH score improves the accuracy of predicting survival of the MDACC score only model (p = 0.006) or RMH score only model (p = 0.0037). The fact that gender was selected as an independent prognostic factor was surprising, although the reason is not clear, it may have to do with the fact that men and women have significant difference in body composition. Limitations of this study include the lack of prospective validation of our findings, and the reliance on a definition of sarcopenia derived from an obese population with cancer, which prevented us to find that SMI as a dichotomised variable was an independent prognostic factor. In summary, our study shows that in the early clinical trials setting, high MDACC score or RMH score >2 are associated with high 90-day mortality. SMI, gender and the MDACC score10 or RMH score are independent prognostic factors for survival, and the addition of
SMI and gender to MDACC10 or RMH scores improves accuracy in predicting overall survival. Overall, our data suggest that body composition plays an important prognostic role for patients with advanced cancer referred for early clinical trials. Because of the important role of prognostic factors in selecting patients for these trials, these observations merit prospective validation. Conflict of interest statement None declared. Acknowledgements The authors acknowledge Joann Aaron, MA, for editing this paper, Bettzy Stephens for data collection and Nina Esfandiari for imaging analyses. This work has no specific funding. References 1. Horstmann E, McCabe MS, Grochow L, et al. Risks and benefits of phase 1 oncology trials, 1991 through 2002. N Engl J Med 2005;352:895–904. 2. Kurzrock R, Benjamin RS. Risks and benefits of phase 1 oncology trials, revisited. N Engl J Med 2005;352:930–2. 3. Janku F, Tsimberidou AM, Garrido-Laguna I, et al. PIK3CA mutations in patients with advanced cancers treated with PI3K/ AKT/mTOR axis inhibitors. Mol Cancer Ther 2011;10:558–65. 4. Tsimberidou AM, Iskander NG, Hong D, et al. Personalized Medicine in a Phase I Clinical Trials Program: The MD Anderson Cancer Center Initiative. Clin Cancer Res 2012;18:6373. 5. Kwak EL, Bang YJ, Camidge DR, et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. N Engl J Med 2010;363:1693–703. 6. Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011;364:2507–16. 7. Arkenau HT, Barriuso J, Olmos D, et al. Prospective validation of a prognostic score to improve patient selection for oncology phase I trials. J Clin Oncol 2009;27:2692–6. 8. Olmos D, A’Hern RP, Marsoni S, et al. Patient selection for oncology phase I trials: a multi-institutional study of prognostic factors. J Clin Oncol 2012;30:996–1004. 9. Garrido-Laguna I, Janku F, Vaklavas C, et al. Validation of the royal marsden hospital prognostic score in patients treated in the phase I clinical trials program at the MD Anderson Cancer Center. Cancer 2012;118:1422–8. 10. Wheler J, Tsimberidou AM, Hong D, et al. Survival of 1,181 Patients in a Phase I Clinic: The MD Anderson Clinical Center for Targeted Therapy Experience. Clin Cancer Res 2012;18:2922–9. 11. Murthy NS, Mukherjee S, Ray G, et al. Dietary factors and cancer chemoprevention: an overview of obesity-related malignancies. J Postgrad Med 2009;55:45–54. 12. Prado CM, Baracos VE, McCargar LJ, et al. Sarcopenia as a determinant of chemotherapy toxicity and time to tumor progression in metastatic breast cancer patients receiving capecitabine treatment. Clin Cancer Res 2009;15:2920–6. 13. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J Epidemiol 1998;147:755–63. 14. Tan BH, Fearon KC. Cachexia: prevalence and impact in medicine. Curr Opin Clin Nutr Metab Care 2008;11:400–7.
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 15. Parsons HA, Baracos VE, Dhillon N, et al. Body composition, symptoms, and survival in advanced cancer patients referred to a phase I service. PLoS One 2012;7:e29330. 16. Tandon P, Ney M, Irwin I, et al. Severe muscle depletion in patients on the liver transplant wait list - its prevalence and independent prognostic value. Liver Transpl 2012;18:1209–16. 17. Montano-Loza AJ, Meza-Junco J, Prado CM, et al. Muscle wasting is associated with mortality in patients with cirrhosis. Clin Gastroenterol Hepatol 2012;10:166–73, 173e1. 18. Lieffers JR, Bathe OF, Fassbender K, et al. Sarcopenia is associated with postoperative infection and delayed recovery from colorectal cancer resection surgery. Br J Cancer 2012;107:931–6. 19. Antoun S, Baracos VE, Birdsell L, et al. Low body mass index and sarcopenia associated with dose-limiting toxicity of sorafenib in patients with renal cell carcinoma. Ann Oncol 2010;21:1594–8. 20. Mourtzakis M, Prado CM, Lieffers JR, et al. A practical and precise approach to quantification of body composition in cancer patients using computed tomography images acquired during routine care. Appl Physiol Nutr Metab 2008;33:997–1006. 21. Shen W, Punyanitya M, Wang Z, et al. Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image. J Appl Physiol 2004;97:2333–8. 22. Shen W, Punyanitya M, Wang Z, et al. Visceral adipose tissue: relations between single-slice areas and total volume. Am J Clin Nutr 2004;80:271–8. 23. Mitsiopoulos N, Baumgartner RN, Heymsfield SB, et al. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol 1998;85:115–22. 24. Heymsfield SB, Smith R, Aulet M, et al. Appendicular skeletal muscle mass: measurement by dual-photon absorptiometry. Am J Clin Nutr 1990;52:214–8. 25. Ladoire S, Bonnetain F, Gauthier M, et al. Visceral fat area as a new independent predictive factor of survival in patients with metastatic renal cell carcinoma treated with antiangiogenic agents. Oncologist 2011;16:71–81. 26. Guiu B, Petit JM, Bonnetain F, et al. Visceral fat area is an independent predictive biomarker of outcome after first-line bevacizumab-based treatment in metastatic colorectal cancer. Gut 2010;59:341–7. 27. Nemesure B, Wu SY, Hennis A, et al. Central adiposity and prostate cancer in a black population. Cancer Epidemiol Biomarkers Prev 2012;21:851–8. 28. Prado CM, Lieffers JR, McCargar LJ, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol 2008;9:629–35.
3075
29. Harrell FE, Lee KL, Mark DB. Tutorial in biostatistics: multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996:361–87. 30. Chau NG, Florescu A, Chan KK, et al. Early mortality and overall survival in oncology phase I trial participants: can we improve patient selection? BMC Cancer 2011;11:426. 31. Tan BH, Birdsell LA, Martin L, et al. Sarcopenia in an overweight or obese patient is an adverse prognostic factor in pancreatic cancer. Clin Cancer Res 2009;15:6973–9. 32. Antoun S, Birdsell L, Sawyer MB, et al. Association of skeletal muscle wasting with treatment with sorafenib in patients with advanced renal cell carcinoma: results from a placebo-controlled study. J Clin Oncol 2010;28:1054–60. 33. Slaviero KA, Read JA, Clarke SJ, et al. Baseline nutritional assessment in advanced cancer patients receiving palliative chemotherapy. Nutr Cancer 2003;46:148–57. 34. Thoresen L, Frykholm G, Lydersen S, et al. Nutritional status, cachexia and survival in patients with advanced colorectal carcinoma. Different assessment criteria for nutritional status provide unequal results. Clin Nutr 2013;32:65–72. 35. Ono T, Guthold R, Strong K. WHO global comparable estimates (ed), Available from http://www.who.int/infobase IBRRef:199999; 2005. 36. Dalal S, Hui D, Bidaut L, et al. Relationships among body mass index, longitudinal body composition alterations, and survival in patients with locally advanced pancreatic cancer receiving chemoradiation: a pilot study. J Pain Symptom Manage 2012;44:181–91. 37. Bachmann J, Heiligensetzer M, Krakowski-Roosen H, et al. Cachexia worsens prognosis in patients with resectable pancreatic cancer. J Gastrointest Surg 2008;12:1193–201. 38. Dodson S, Baracos VE, Jatoi A, et al. Muscle wasting in cancer cachexia: clinical implications, diagnosis, and emerging treatment strategies. Annu Rev Med 2011;62:265–79. 39. Kohara K, Ochi M, Tabara Y, et al. Leptin in sarcopenic visceral obesity: possible link between adipocytes and myocytes. PLoS One 2011;6:e24633. 40. Peng PD, van Vledder MG, Tsai S, et al. Sarcopenia negatively impacts short-term outcomes in patients undergoing hepatic resection for colorectal liver metastasis. HPB (Oxford) 2011;13:439–46. 41. van Vledder MG, Levolger S, Ayez N, et al. Body composition and outcome in patients undergoing resection of colorectal liver metastases. Br J Surg 2012;99:550–7. 42. Sabel MS, Lee J, Cai S, et al. Sarcopenia as a prognostic factor among patients with stage III melanoma. Ann Surg Oncol 2011;18:3579–85.
Available at www.sciencedirect.com
journal homepage: www.ejcancer.com
Body composition and survival in the early clinical trials setting H. Veasey Rodrigues a,⇑, V.E. Baracos b, J.J. Wheler a, H.A. Parsons a, D.S. Hong a, A. Naing a, S. Fu a, G. Falchoock a, A.M. Tsimberidou a, S. Piha-Paul a, G. Chisholm c, R. Kurzrock a a
Department of Investigational Cancer Therapeutics (Phase I Clinical Trials Program), The University of Texas MD Anderson Cancer Center, Houston, TX, USA b Department of Oncology, University of Alberta, Edmonton, Canada c Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA
Available online 15 July 2013
KEYWORDS Body composition Sarcopenia Prognostic score Phase I clinical trial
Abstract Purpose: Delineate the relationships between body composition parameters, 90-day mortality and overall survival, and correlate them with known prognostic factors in an early clinical trials clinic. Patients and methods: We studied 306 consecutive patients with various tumours; body composition was analysed by computerised tomography images. Survival was measured from the first clinic visit, at 90-day period and until death/last follow-up visit. Results: Median patient age was 56 years; 159 patients were men. Ninety-day mortality rate was 12%. Median overall survival was 9 months. In multivariate analyses, high MD Anderson Cancer Center (MDACC) score (p < 0.0001) [lactate dehydrogenase (LDH) > normal, albumin < normal, Eastern Cooperative Oncology Group (ECOG) performance status > 1, metastatic sites > 2, gastrointestinal (GI) tumours], low skeletal muscle index (SMI) (p = 0.0406) and male gender (p = 0.0077) were independent predictors of poor survival. If Royal Marsden Hospital (RMH) score (LDH > normal, albumin < normal, metastatic sites > 2) was used in lieu of MDACC score, it was also significant (p = 0.0003). Including SMI and gender in the MDACC or RMH score improved the accuracy of the original model (p = 0.006 and p = 0.0037, respectively). Conclusion: Patients with low SMI have shorter survival. Gender and SMI strengthens the accuracy of MDACC or RMH scores as prognostic tools. Prospective validation of these findings is warranted. Ó 2013 Elsevier Ltd. All rights reserved.
⇑ Corresponding author: Address: Av. Europa, 105, Sa˜o Paulo, SP 01421-001, Brazil. Tel./fax: +55 11 30675400.
E-mail address: [email protected] (H. Veasey Rodrigues). 0959-8049/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejca.2013.06.026
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
3069
1. Introduction
2. Patients and methods
Early clinical trials are designed to determine metabolic and pharmacologic effects of drugs in humans, side-effects associated with increasing doses, maximum tolerated dose, and early evidence of efficacy and response. Patients referred to phase I clinical trials usually present with treatment-refractory, advanced malignancies, and many will not benefit from experimental treatments.1,2 Therefore, understanding prognostic factors can assist in determining possible outcomes. Personalised therapy that is, matching patients with agents based on an individualised approach is playing an increasingly important role in cancer therapy.3–6 The significance of host factors such as body composition to this field has not been well studied. Efforts have been made to delineate objective and reliable prognostic parameters. The Royal Marsden Hospital (RMH) has developed a prognostic scoring system using objectively measured parameters [>2 sites of metastasis, albumin below normal and lactate dehydrogenase (LDH) above normal] that have been validated in multi-institutional studies,7,8 including the phase I clinic at MD Anderson.9 Our group has also studied other prognostic factors related to survival in the phase I patient population. Among these are the gastrointestinal (GI) tumour type and poor Eastern Cooperative Oncology Group (ECOG) performance status and, when grouped with the RMH variables, the MD Anderson Cancer Center (MDACC) score10 may strengthen the ability of the RMH score to predict a poor outcome. Body weight and body composition (fat and lean mass) play a crucial role in the aetiology, prognosis and treatment outcomes of cancer.11,12 Sarcopenia, defined as absolute muscle mass below 2 or more standard deviations of the muscle mass in healthy young adults13 is closely associated with cancer cachexia, and has a significant impact on the prognosis of cancer and other chronic diseases.14–19 Patients referred to clinical trials usually present with progressive disease despite conventional management. Therefore, they are particularly susceptible to the end-organ effects of cachexia, reflected as changes in body weight and body composition. Clearly, in view of these factors, understanding how body weight and composition impact outcomes and survival in this population is important. In this study, we analysed body weight and body composition in 306 patients who were referred to the Clinical Center for Targeted Therapy (phase I clinic) of the MD Anderson Cancer Center. Our goal was to delineate the relationships between body composition parameters, including sarcopenia, 90-day mortality and overall survival, and correlate them with known prognostic factors.
A total of 306 consecutive patients with various advanced cancers referred to the phase I clinic starting in December 2004, who met protocol criteria for inclusion, were analysed. These patients were part of a larger pool of patients from a previous study that evaluated survival of patients in the phase I setting.10 Patients were included if the computed tomography (CT) imaging test had taken place within 5–50 days of their first visit to the phase I clinic. Images were used to assess body composition. All patients received treatment on early clinical trials including cytotoxic and/or targeted agents, systemic and/or local-regional therapy (intra-hepatic arterial infusion). The study was conducted according to guidelines of the Institutional Review Board at MD Anderson, and written informed consents for all investigational treatments were obtained. Patients were followed up to determine length of survival and relevant variables, including body composition. 2.1. Demographic and laboratory data Age, gender, cancer diagnosis, height and weight data were collected from patients’ electronic medical records (Table 1). Weight measurements were culled from the day closest to the date of CT imaging at the first visit to the phase I clinic; median time from weight to CT imaging was 1 day (range, 10 days before to 20 days after imaging). Laboratory values were collected from the closest day to the first visit to the clinic; median time from laboratory tests to first visit was 0 days (range, 15 days before to 15 days after the first visit). The median time between CT imaging and first clinic visit was 13 days (range, 0–50 days before visit), only 8.2% (25/ 306) of the patients had imaging done between 31 and 50 days prior to clinic visit. 2.2. Survival data Overall survival was measured from the day of the first visit to the phase I clinic until death or final follow-up visit. Patients still alive at last follow-up were censored at that date. Ninety-day survival was calculated by censoring alive patients at day 90. 2.3. Body composition assessments Body mass index (BMI) was calculated dividing patient’s weight (kg) by height (m2). Body composition (muscle and fat body area) was estimated using the validated method described below. Routine abdominal CT images at the level of the 3rd lumbar vertebra (L3) were chosen for analysis. The use
3070
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 Table 1 Patient characteristics. Women (147)
Men (159)
Total (306)
Age, years (range)
55 (16–76)
60 (24–84)
56 (16–84)
ECOG (%) 0 1 2 3 4 NAa
36 (24) 92 (63) 17 (12) 1 (0.5) 0 1 (0.5)
51 (32) 95 (60) 8 (5) 2 (1) 0 3 (2)
87 (29) 187 (61) 25 (8) 3 (1) 0 4 (1)
66.9 (28.9–150.5)
86.0 (52.5–161.3)
78.5 (28.9–161.3)
13 58 44 32
1 (1) 47 (29.5) 64 (40) 47 (29.5)
14 (5) 105 (34) 108 (35) 79 (26)
Body composition parameters, cm2/m2 (95% CI) Skeletal muscle index 40.9 (40.3–42.6) Adipose index 64.4 (63.7–83.4)
50.5 (49.4–52.6) 91.8 (83.8–103.6)
44.9 (45.7–47.3) 79.7 (77.0–91.2)
Sarcopenia (%) NAb
51 (35) 1
93 (59) 1
144 (47)
Diagnosis (%) Gastrointestinal Genitourinary Breast Thyroid Melanoma Head and neck Lung Gynaecological Sarcoma Othersc
47 (32) 2 (1) 30 (20) 16 (11) 7 (5) 5 (3) 10 (7) 14 (10) 5 (3) 11 (8)
52 (33) 33 (21) 1 (1) 14 (9) 16 (10) 15 (9) 8 (5) – 7 (4) 13 (8)
99 35 31 30 23 20 18 14 12 24
RMH prognostic score 0 1 2 3 NAa
48 (33) 64 (44) 27 (18) 7 (5) 1
71 (45) 56 (35) 28 (18) 4 (2) 0
119 (39) 120 (39) 55 (18) 11 (4) 1
MDACC prognostic score 0 1 2 3 4 5 NAa
8 (5) 47 (32) 44 (30) 30 (21) 14 (10) 3 (2) 1
23 (14) 47 (30) 45 (28) 31 (20) 11 (7) 2 (1) 0
31 (10) 94 (31) 89 (29) 61 (20) 25 (8) 5 (2) 1
No of trials 1–2 >2
139 (95) 8 (5)
143 (90) 16 (10)
282 (92) 24 (8)
Weight, kg 2
BMI, kg/m (%) <18.5 18.5–24.9 25–29.9 P30
(9) (39) (30) (22)
(32) (11) (10) (9) (8) (7) (6) (5) (4) (8)
BMI: body mass index; <18.5 kg/m2 = underweight, 18.5–24.9 kg/m2 = normal, 25–29.9 kg/ m2 = overweight, P30 kg/m2 = obese. ECOG: Eastern Cooperative Oncology Group performance status score. MDACC: MD Anderson Cancer Center. NA: not available. RMH: Royal Marsden Hospital. a ECOG performance status not documented at the first visit to phase I clinic. b Not available due to technical limitation on CT imaging. c Others (non-melanoma skin cancer, thymus, unknown primary, peritoneal, CNS, lymphoma).
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
of L3 as the landmark for body composition analysis has previously been described and validated against dual X-ray absorptiometry in healthy populations and in patients with advanced cancer.20–22 Muscle, intramuscular fat, subcutaneous fat and visceral fat were identified by a single assessor trained in the specific anatomy of these tissues, demarcated using previously described Hounsfield unit thresholds23,24 and quantified in crosssectional areas (cm2) by the SliceOMatic software, version 4.3 (Tomovision, Montreal, QC, Canada). Fat tissue analysis was done in its totality and also dividing into subcutaneous fat and visceral fat because there is evidence that visceral fat may be associated with worse outcome.25–27 Sarcopenia was defined by a lumbar skeletal muscle index (skeletal muscle area at L3 divided by the height squared) lower than 38.5 cm2/m2 for women and lower than 52.4 cm2/m2 for men, as previously described.28 Adipose index was determined by dividing adipose tissue area at L3 by the height squared. 2.4. End-points and statistical analyses The goal of this study was to investigate the impact of body composition parameters, including sarcopenia, as prognostic factors for 90-day mortality and overall survival in patients referred to early clinical trials, and compare results with prognostic scores. Descriptive statistics were used to summarise patients’ characteristics at baseline by gender. Differences in categorical variables were determined by chi-square and Fisher’s exact tests, when applicable. Differences in continuous variables were determined using a t-test. Cox regression was performed for 90-day and overall survival. Each variable was first tested at a 10% significance level in a univariate model. The significant variables were then tested in a ‘univariate’ model adjusted for age and gender. Variables in the adjusted univariate model, which were significant at the 10% level, were then tested together in multivariate analysis. Harrel’s C statistic29 was calculated for multivariate models and z-tests were performed to test relative predictive accuracy. Analyses were performed using SAS 9.2 (SAS Institute, Cary NC) and R (The R Project for Statistical Computing, Vienna, Austria). 3. Results 3.1. Patient characteristics Fifty-two percent of patients were men (159/306); median age was 56 years; and ECOG performance status was 0 or 1 for 90% of patients (274/305) (one patient did not have ECOG status recorded). Sixty-one percent (187/306) of patients were overweight or obese, 34% (105/306) were normal weight and 5% (14/306) were
3071
underweight. Sarcopenia was found in 47% (144/304) of patients (two patients had technical issues surrounding CT imaging and sarcopenia could not be evaluated, though fat area was evaluated). The most frequent diagnoses were GI (32%, 99/306), genitourinary (11%, 35/ 306) and breast cancers (10%, 31/306). Among GI tumours, colorectal represented the vast majority (69%, 68/99), followed by upper GI (13%, 13/99), pancreatic (11%, 11/99) and small intestine tumours (7%, 7/99). The median number of prior therapies was four (range, 0–12). Only less than half percent (13/306) were previously exposed to tyrosine kinase inhibitors such as sorafenib. Patients with good prognosis using RMH score (0–1) were 78% (239/306) and 70% (214/306) had a low-to-intermediate (0–2) MDACC risk score10 (Table 1). 3.2. Clinical trials Patients were treated in 77 clinical trials, 92% (282/ 306) of patients participated in two or fewer trials (Table 1). Most of the trials consisted of targeted drugs, either as single agents (56%, 43/77) or in combination (13%, 10/77). The remainder trials consisted of combinations of cytotoxic and targeted drugs (15%, 12/77), single cytotoxic agents (12%, 9/77) and combinations of cytotoxic agents (4%, 3/77). 3.3. 90-Day mortality The 90-day mortality rate was 12% (36/306). The following categorical variables were evaluated as predictors of 90-day mortality: gender (female versus male), RMH score (P2 versus <2), albumin (<3.5 g/dL versus P3.5 g/dL), LDH (>618 IU/L versus 6618 IU/L), GI tumours (yes versus no), ECOG status (P1 versus 0), number of metastatic sites (P3 versus <3) and sarcopenia (yes versus no). The following continuous variables were evaluated as predictors of 90-day mortality: age, BMI, weight, MDACC score, skeletal muscle area (SMA), skeletal muscle index (SMI), visceral adipose area (VAA), subcutaneous adipose area (SCAA), total adipose area (TAA), adipose index (AI). 3.3.1. Univariate analysis Albumin <3.5 g/dL (p = 0.0163), low BMI (p = 0.0002), GI tumour (p = 0.0442), LDH > 618 IU/L (p = 0.0139), high MDACC score (p = 0.0001), metastatic sites P3 (p = 0.0136), RMH score P2 (p = 0.0022), low SCAA (p = 0.0146), low TAA (p = 0.0431) and low weight (p = 0.0014) were associated with a higher 90-day mortality (Table 2). These values were adjusted to age and gender, and adjusted TAA was not significant (p = 0.0793), whereas all variables mentioned above remained significant predictors for 90-day mortality. Because low SMI showed a trend
3072
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
Table 2 Univariate and multivariate analysis of predictor variables of 90-day mortality. Variables
HR
p Value
Univariate model Age* AI* (cm2/m2) Albumin (<3.5 g/dL versus >3.5 g/dL) BMI* (kg/m2) ECOG status (P1 versus <1) Gender (female versus male) GI tumour (Y versus N) LDH (>618 IU/L versus <618 IU/L) MDACC score* Metastatic sites (P3 versus <3) RMH score (P2 versus <2) Sarcopenia (Y versus N) SCAA* (cm2) SMA* (cm2) SMI* (cm2/m2) TAA* (cm2) VAA* (cm2) Weight* (kg)
0.982 0.997 2.521 0.874 1.750 1.121 1.958 2.273 1.729 2.302 2.819 1.678 0.995 0.993 0.966 0.998 0.997 0.969
0.1523 0.3276 0.0163 0.0002 0.1839 0.7322 0.0442 0.0139 <0.0001 0.0136 0.0022 0.1338 0.0146 0.1511 0.0793 0.0431 0.1147 0.0014
Multivariate model – Composite MDACC score 0.976 Age* BMI* 0.939 Gender (female versus male) 0.734 1.794 MDACC score* SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.996
0.0816 0.5194 0.4679 0.0002 0.7455 0.8921
Multivariable model – MDACC score individual variables 0.976 Age* Albumin (<3.5 g/dL versus > 3.5 g/dL) 1.913 BMI* (kg/m2) 0.929 Gender (female versus male) 0.731 GI tumour (Y versus N) 2.055 LDH (>618 IU/L versus <618 IU/L) 2.230 Metastatic sites (P3 versus <3) 1.612 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.992
0.0840 0.1371 0.4552 0.4656 0.0530 0.0222 0.2023 0.8034 0.8078
AI, adipose index; BMI, Body mass index; ECOG, Eastern Cooperative Oncology Group performance status score; GI, gastrointestinal; LDH, lactate dehydrogenase; MDACC, MD Anderson Cancer Center; SCAA, subcutaneous adipose area; SMA, skeletal muscle area; SMI, skeletal muscle index; TAA, total adipose area; VAA, visceral adipose area. * Continuous variables.
(p = 0.0793) to higher 90-day mortality, it was included in the adjusted univariate model and was found to be a significant predictor (p = 0.0338). 3.3.2. Multivariate analysis The multivariable analysis was performed with the significant variables from adjusted univariate model avoiding component variables (e.g. weight is a component of BMI). A high MDACC score was the only variable associated with increased 90-day mortality (p = 0.0002). If the analysis was done using the individual components of the MDACC score, only LDH > 618 IU/L (p = 0.0222) was significant. GI tumours showed a trend towards significance as a predictor of 90-day mortality (p = 0.0530) (Table 2).
If RMH score was used instead of MDACC score, it was also selected as predictor of 90-day mortality (p = 0.0126) along with GI tumour (p = 0.0271). 3.4. Overall survival The median overall survival was 9.0 months (IC 95% 7.2–10.8). The same variables evaluated as predictors of 90-day mortality were evaluated for overall survival. 3.4.1. Univariate analysis The univariate model showed that albumin <3.5 g/dL (p = 0.0092), ECOG P1 (p = 0.030), GI tumour (p = 0.0166), LDH >618 IU/L (p < 0.0001), high MDACC score (p < 0.0001), metastatic sites P3 (p = 0.001), RMH score P2 (p < 0.0001), sarcopenia (p = 0.0138), low SCAA (p = 0.0362) and low SMI (p = 0.0359) were associated with a shorter survival. These variables all remained significant after adjustment for age and gender. We have also analysed the overall survival according to body mass index and low skeletal muscle (sarcopenia). We found that there is a trend (p = 0.06) to a better survival for no sarcopenic patients, especially if BMI > 25 kg/m2 (11.5 months, 95% CI 9.1– 13.9), if BMI < 25 kg/m2 median survival was 8.6 months (95% CI 7.1–10.1). Sarcopenic patients had a median survival of 7.8 (95% CI 5.0–10.6) and 7.0 months (95% CI, 5.2–8.7) if BMI > 25 kg/m2 and BMI 6 25 kg/m2, respectively. 3.4.2. Multivariate analysis The multivariate analysis was performed in different models to avoid having component variables (e.g. the individual variables of the MDACC score) analysed together. High MDACC score (p < 0.0001), low SMI (p = 0.0406) and male gender (p = 0.0077) were independently associated with shorter survival (Table 3). When the MDACC score components were analysed individually, male gender (p = 0.0071), LDH > 618 IU/ L (p < 0.0001) and low SMI (p = 0.0445) were associated with shorter survival. ECOG performance status P1 showed a trend to shorter survival (p = 0.0617) (Table 3). Sarcopenia as a dichotomous variable (SMI < 52.4cm2/m2 for men and <38.5cm2/m2 for women) was not selected as an independent variable. The RMH score was not included in the multivariate analysis because it is included in the MDACC score, but RMH score (p = 0.0003) is also selected as predictor of overall survival if used instead of the MDACC score in the multivariate model, along with low SMI (p = 0.0300) and male gender (p = 0.0093). 3.4.3. Including SMI in composite models to predict survival SMI was included in the composite scores to investigate whether this variable increased the accuracy of the
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 Table 3 Univariate and multivariate analysis of predictor variables of overall survival. Variables
HR
p value
Univariate model Age* AI* (cm2/m2) Albumin (<3.5 g/dL versus >3.5 g/dL) BMI* (kg/m2) ECOG status (P1 versus <1) Gender (female versus male) GI tumour (Y versus N) LDH (>618 IU/L versus <618 IU/L) MDACC score* Metastatic sites (P3 versus <3) RMH score (P2 versus <2) Sarcopenia (Y versus N) SCAA* (cm2) SMA* (cm2) SMI* (cm2/m2) TAA* (cm2) VAA* (cm2) Weight* (kg)
1.000 0.999 1.579 0.983 1.332 0.861 1.357 2.016 1.363 1.491 2.006 1.340 0.999 0.998 0.987 0.999 0.999 0.997
0.963 0.264 0.0092 0.101 0.030 0.203 0.0166 <0.0001 <0.0001 0.001 <0.0001 0.0138 0.036 0.194 0.036 0.159 0.296 0.340
Multivariable model – Composite MDACC score 0.993 Age* Gender (female versus male) 0.634 MDACC score* 1.390 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.982
0.1525 0.0077 <0.0001 0.4365 0.0406
Multivariable model – MDACC score individual variables 0.992 Age* Albumin (<3.5 versus P3.5) 1.128 ECOG (>1 versus 0) 1.319 Gender (female versus male) 0.626 GI tumour (Y versus N) 1.218 LDH (>618 versus 6618) 2.142 Metastatic Sites (P3 versus <3) 1.238 SCAA* (cm2) 0.999 SMI* (cm2/m2) 0.981
0.1285 0.5572 0.0617 0.0071 0.1838 <0.0001 0.1518 0.1151 0.0445
AI, adipose index; BMI, Body mass index; ECOG, Eastern Cooperative Oncology Group performance status score; GI, gastrointestinal; LDH, lactate dehydrogenase; MDACC, MD Anderson Cancer Center; SCAA, subcutaneous adipose area; SMA, skeletal muscle area; SMI, skeletal muscle index; TAA, total adipose area; VAA, visceral adipose area. * Continuous variable.
composite model. A new model including MDACC score, gender and SMI has a Harrel’s C statistic of 0.62 and the MDACC score (only) model has a Harrel’s C statistic of 0.59. Harrel’s C statistic measures the general predictive power of a survival model.29 Thus, the new model has a higher Harrel’s C statistic and the difference is statistically significant at p = 0.006, indicating that the new model improves the predictive accuracy of the original MDACC score (only) model. The same is true for RMH score; the addition of SMI and gender improved the accuracy of the model compared to the RMH score (only) model (0.62 versus 0.56, p = 0.0037). SMI, however, did not improve the score when gender was excluded from the model.
3073
4. Discussion Patient selection in early anticancer drug development is important for several reasons, including to avoid harm and medical futility because of uncertainty regarding drug toxicity and activity, and avoid confounding factors that may negatively impact the drug safety (adverse event) profile and efficacy data. Several groups, including ours, have described composite scores based on objective parameters that can be used to guide therapy decision-making.7,8,10,30 Survival prognostic scores are also helpful for identifying patients at risk for early death (90-day mortality). Although patients referred to early clinical trials have advanced cancers, a large proportion of enrolled patients had favourable RMH prognostic scores (56– 79%).7,9 Seventy-eight percent of our patients had good RMH prognostic scores and 70% had low-to-intermediate MDACC risk scores.10 This result might be explained by referral bias, because it is known that clinical trials generally exclude severely ill patients, and/or the increased availability of clinical trials in tertiary cancer centres. Another aspect of a referral bias can explain the unusual prevalence of cancer subtypes in our population. Lung cancer was less frequent than thyroid cancer and melanoma for example, and may have influenced our findings, since some cancers are more related to cachexia than others. Sarcopenia was found in 47% of patients in concordance with our previous finding of 51% of sarcopenia in patients seen at MDACC-Phase I clinic.15 Other authors reported sarcopenia in 51% of patients with pancreatic cancer31 and 52.5% in patients with renal cancer.7 It has been well documented that exposure to sorafenib, a tyrosine kinase inhibitor, can promote muscle loss/sarcopenia,32 however, few of our patients (13/ 306) were previously treated with sorafenib or other VEGFR tyrosine kinase inhibitors. The majority (61%) our population was overweight or obese, a percentage consistent with other reports7,15; this is a frequent finding in advanced cancer patients,33,34 particularly those from developed nations where overweight and obesity are epidemic.35 The 90-day mortality in this study was evaluated from the day of the first visit to our clinic, and not the first day of the clinical trial on which a patient was enrolled, as reported by other groups.8,30 This could account for our slightly lower 90-day mortality, which was 12% in the current study versus the 15–20% reported previously.7,8,30 A high MDACC score10 was associated with increased 90-day mortality in multivariate analysis (Table 2). Body weight, BMI and other body composition parameters were not associated with increased 90-day mortality in our cohort when MDACC as a composite score was included. The individual component analysis showed that only LDH above normal
3074
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075
was independently associated with increased 90-day mortality. If MDACC was substituted for RMH score it was also selected as an independent variable predicting 90-day survival, along with GI tumour type. Until now, the known prognostic scores (MDACC and RMH) have not considered body weight and body composition parameters as prognostic variables, although scientific evidence links cachexia–sarcopenia and sarcopenic obesity with cancer prognosis and survival.15,28,31,36–38 Interestingly, Kohara et al.39 demonstrated a relation between sarcopenic obesity and high levels of leptin, which is a pro-angiogenic factor, and could explain in part the inferior outcome of this group. This study demonstrated that low skeletal muscle index (SMI) as a continuous variable, was associated with shorter overall survival in a multivariate analysis model (Table 3), and it was an independent predictor of overall survival even in the presence of validated prognostic scores (MDACC and RMH) and other individual factors, such as LDH and ECOG performance status. Sarcopenia, which is the SMI variable dichotomised, was, however, not an independent factor predicting shorter survival in a multivariable model. The latter may be because the parameters defining the cutoff for sarcopenia are derived from an obese population,28 and did not include the full spectrum of cancer patients. The literature demonstrates that low skeletal muscle mass/sarcopenia is frequently associated with poor outcomes. For instance, in patients with colorectal cancer undergoing resection of liver metastasis, sarcopenia was associated with greater postoperative complications40 and worse progression-free survival and overall survival.41 In overweight or obese cancer patients, sarcopenia is an independent predictor of survival,15,28,31 and in patients with stage III melanoma sarcopenia was an independent factor for disease-free survival.42 Finally, our study demonstrates that addition of SMI and gender in a model with MDACC score10 or RMH score improves the accuracy of predicting survival of the MDACC score only model (p = 0.006) or RMH score only model (p = 0.0037). The fact that gender was selected as an independent prognostic factor was surprising, although the reason is not clear, it may have to do with the fact that men and women have significant difference in body composition. Limitations of this study include the lack of prospective validation of our findings, and the reliance on a definition of sarcopenia derived from an obese population with cancer, which prevented us to find that SMI as a dichotomised variable was an independent prognostic factor. In summary, our study shows that in the early clinical trials setting, high MDACC score or RMH score >2 are associated with high 90-day mortality. SMI, gender and the MDACC score10 or RMH score are independent prognostic factors for survival, and the addition of
SMI and gender to MDACC10 or RMH scores improves accuracy in predicting overall survival. Overall, our data suggest that body composition plays an important prognostic role for patients with advanced cancer referred for early clinical trials. Because of the important role of prognostic factors in selecting patients for these trials, these observations merit prospective validation. Conflict of interest statement None declared. Acknowledgements The authors acknowledge Joann Aaron, MA, for editing this paper, Bettzy Stephens for data collection and Nina Esfandiari for imaging analyses. This work has no specific funding. References 1. Horstmann E, McCabe MS, Grochow L, et al. Risks and benefits of phase 1 oncology trials, 1991 through 2002. N Engl J Med 2005;352:895–904. 2. Kurzrock R, Benjamin RS. Risks and benefits of phase 1 oncology trials, revisited. N Engl J Med 2005;352:930–2. 3. Janku F, Tsimberidou AM, Garrido-Laguna I, et al. PIK3CA mutations in patients with advanced cancers treated with PI3K/ AKT/mTOR axis inhibitors. Mol Cancer Ther 2011;10:558–65. 4. Tsimberidou AM, Iskander NG, Hong D, et al. Personalized Medicine in a Phase I Clinical Trials Program: The MD Anderson Cancer Center Initiative. Clin Cancer Res 2012;18:6373. 5. Kwak EL, Bang YJ, Camidge DR, et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. N Engl J Med 2010;363:1693–703. 6. Chapman PB, Hauschild A, Robert C, et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011;364:2507–16. 7. Arkenau HT, Barriuso J, Olmos D, et al. Prospective validation of a prognostic score to improve patient selection for oncology phase I trials. J Clin Oncol 2009;27:2692–6. 8. Olmos D, A’Hern RP, Marsoni S, et al. Patient selection for oncology phase I trials: a multi-institutional study of prognostic factors. J Clin Oncol 2012;30:996–1004. 9. Garrido-Laguna I, Janku F, Vaklavas C, et al. Validation of the royal marsden hospital prognostic score in patients treated in the phase I clinical trials program at the MD Anderson Cancer Center. Cancer 2012;118:1422–8. 10. Wheler J, Tsimberidou AM, Hong D, et al. Survival of 1,181 Patients in a Phase I Clinic: The MD Anderson Clinical Center for Targeted Therapy Experience. Clin Cancer Res 2012;18:2922–9. 11. Murthy NS, Mukherjee S, Ray G, et al. Dietary factors and cancer chemoprevention: an overview of obesity-related malignancies. J Postgrad Med 2009;55:45–54. 12. Prado CM, Baracos VE, McCargar LJ, et al. Sarcopenia as a determinant of chemotherapy toxicity and time to tumor progression in metastatic breast cancer patients receiving capecitabine treatment. Clin Cancer Res 2009;15:2920–6. 13. Baumgartner RN, Koehler KM, Gallagher D, et al. Epidemiology of sarcopenia among the elderly in New Mexico. Am J Epidemiol 1998;147:755–63. 14. Tan BH, Fearon KC. Cachexia: prevalence and impact in medicine. Curr Opin Clin Nutr Metab Care 2008;11:400–7.
H. Veasey Rodrigues et al. / European Journal of Cancer 49 (2013) 3068–3075 15. Parsons HA, Baracos VE, Dhillon N, et al. Body composition, symptoms, and survival in advanced cancer patients referred to a phase I service. PLoS One 2012;7:e29330. 16. Tandon P, Ney M, Irwin I, et al. Severe muscle depletion in patients on the liver transplant wait list - its prevalence and independent prognostic value. Liver Transpl 2012;18:1209–16. 17. Montano-Loza AJ, Meza-Junco J, Prado CM, et al. Muscle wasting is associated with mortality in patients with cirrhosis. Clin Gastroenterol Hepatol 2012;10:166–73, 173e1. 18. Lieffers JR, Bathe OF, Fassbender K, et al. Sarcopenia is associated with postoperative infection and delayed recovery from colorectal cancer resection surgery. Br J Cancer 2012;107:931–6. 19. Antoun S, Baracos VE, Birdsell L, et al. Low body mass index and sarcopenia associated with dose-limiting toxicity of sorafenib in patients with renal cell carcinoma. Ann Oncol 2010;21:1594–8. 20. Mourtzakis M, Prado CM, Lieffers JR, et al. A practical and precise approach to quantification of body composition in cancer patients using computed tomography images acquired during routine care. Appl Physiol Nutr Metab 2008;33:997–1006. 21. Shen W, Punyanitya M, Wang Z, et al. Total body skeletal muscle and adipose tissue volumes: estimation from a single abdominal cross-sectional image. J Appl Physiol 2004;97:2333–8. 22. Shen W, Punyanitya M, Wang Z, et al. Visceral adipose tissue: relations between single-slice areas and total volume. Am J Clin Nutr 2004;80:271–8. 23. Mitsiopoulos N, Baumgartner RN, Heymsfield SB, et al. Cadaver validation of skeletal muscle measurement by magnetic resonance imaging and computerized tomography. J Appl Physiol 1998;85:115–22. 24. Heymsfield SB, Smith R, Aulet M, et al. Appendicular skeletal muscle mass: measurement by dual-photon absorptiometry. Am J Clin Nutr 1990;52:214–8. 25. Ladoire S, Bonnetain F, Gauthier M, et al. Visceral fat area as a new independent predictive factor of survival in patients with metastatic renal cell carcinoma treated with antiangiogenic agents. Oncologist 2011;16:71–81. 26. Guiu B, Petit JM, Bonnetain F, et al. Visceral fat area is an independent predictive biomarker of outcome after first-line bevacizumab-based treatment in metastatic colorectal cancer. Gut 2010;59:341–7. 27. Nemesure B, Wu SY, Hennis A, et al. Central adiposity and prostate cancer in a black population. Cancer Epidemiol Biomarkers Prev 2012;21:851–8. 28. Prado CM, Lieffers JR, McCargar LJ, et al. Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study. Lancet Oncol 2008;9:629–35.
3075
29. Harrell FE, Lee KL, Mark DB. Tutorial in biostatistics: multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996:361–87. 30. Chau NG, Florescu A, Chan KK, et al. Early mortality and overall survival in oncology phase I trial participants: can we improve patient selection? BMC Cancer 2011;11:426. 31. Tan BH, Birdsell LA, Martin L, et al. Sarcopenia in an overweight or obese patient is an adverse prognostic factor in pancreatic cancer. Clin Cancer Res 2009;15:6973–9. 32. Antoun S, Birdsell L, Sawyer MB, et al. Association of skeletal muscle wasting with treatment with sorafenib in patients with advanced renal cell carcinoma: results from a placebo-controlled study. J Clin Oncol 2010;28:1054–60. 33. Slaviero KA, Read JA, Clarke SJ, et al. Baseline nutritional assessment in advanced cancer patients receiving palliative chemotherapy. Nutr Cancer 2003;46:148–57. 34. Thoresen L, Frykholm G, Lydersen S, et al. Nutritional status, cachexia and survival in patients with advanced colorectal carcinoma. Different assessment criteria for nutritional status provide unequal results. Clin Nutr 2013;32:65–72. 35. Ono T, Guthold R, Strong K. WHO global comparable estimates (ed), Available from http://www.who.int/infobase IBRRef:199999; 2005. 36. Dalal S, Hui D, Bidaut L, et al. Relationships among body mass index, longitudinal body composition alterations, and survival in patients with locally advanced pancreatic cancer receiving chemoradiation: a pilot study. J Pain Symptom Manage 2012;44:181–91. 37. Bachmann J, Heiligensetzer M, Krakowski-Roosen H, et al. Cachexia worsens prognosis in patients with resectable pancreatic cancer. J Gastrointest Surg 2008;12:1193–201. 38. Dodson S, Baracos VE, Jatoi A, et al. Muscle wasting in cancer cachexia: clinical implications, diagnosis, and emerging treatment strategies. Annu Rev Med 2011;62:265–79. 39. Kohara K, Ochi M, Tabara Y, et al. Leptin in sarcopenic visceral obesity: possible link between adipocytes and myocytes. PLoS One 2011;6:e24633. 40. Peng PD, van Vledder MG, Tsai S, et al. Sarcopenia negatively impacts short-term outcomes in patients undergoing hepatic resection for colorectal liver metastasis. HPB (Oxford) 2011;13:439–46. 41. van Vledder MG, Levolger S, Ayez N, et al. Body composition and outcome in patients undergoing resection of colorectal liver metastases. Br J Surg 2012;99:550–7. 42. Sabel MS, Lee J, Cai S, et al. Sarcopenia as a prognostic factor among patients with stage III melanoma. Ann Surg Oncol 2011;18:3579–85.