Worsening Central Sarcopenia and Increasing Intra-Abdominal Fat Correlate with Decreased Survival in Patients with Adrenocortical Carcinoma

Barbra S. Miller . Kathleen M. Ignatoski . Stephanie Daignault · Ceit Lindland . Megan Doherty . Paul G. Gauger . Gary D. Hammer . Stewart C. Wang . Gerard M. Doherty . The University of Michigan Analytical Morphomics Group

Published online: 19 April 2012 @ Société Internationale de Chirurgie 2012

Abstract

Background Accurate prediction of survival from adre- nocortical carcinoma (ACC) is difficult and current staging models are unreliable. Central sarcopenia as part of the cachexia syndrome is a marker of frailty and predicts mortality. This study seeks to confirm that psoas muscle density (PMD), lean psoas muscle area (LPMA), lumbar skeletal muscle index (LSMI), and intra-abdominal (IA) or subcutaneous fat (SC) can be used in combination to more accurately predict survival in ACC patients.

Methods PMD, LPMA, IA, and SC fat were measured on serial CT scans of patients with ACC. Clinical outcome was correlated with quantitative data from patients with ACC and analyzed. A linear regression model was used to describe the relationship between PMD, LPMA, LSMI, IA, and SC fat, time to recurrence, and length of survival according to tumor stage.

Results One hundred twenty-five ACC patients (94 females) were treated from 2005 to 2011. Significant morphometric predictors of survival include PMD, LPMA, and IA fat (p ≤0.0001, ≤0.0024, <0.0001, respectively) and improve prediction of survival compared to using stage alone. A 100-mm2 increase in LPMA confers an 8 % lower hazard of death. LSMI does not change significantly between stages (p = 0.3196).

Conclusion Decreased PMD, LPMA, and increased IA fat suggest decreased survival in ACC patients and corre- late with traditional staging systems. A more precise pre- diction of survival may be achieved when staging systems and morphometric measures are used in combination. Serial measurements of morphometric data are possible. The rate of change of these variables over time may be more important than the absolute value.

B. S. Miller () . K. M. Ignatoski . C. Lindland ·

M. Doherty · P. G. Gauger . G. M. Doherty Division of Endocrine Surgery, Section of General Surgery, Department of Surgery, University of Michigan Health System,

2920F Taubman Center, 1500 E. Medical Center Drive, Ann Arbor, MI 48109, USA

e-mail: barbram@umich.edu

S. Daignault

Comprehensive Cancer Center Biostatistics Unit, University of Michigan Health System, Ann Arbor, MI, USA

G. D. Hammer

Division of Metabolism, Endocrinology and Diabetes, University of Michigan Health System, Ann Arbor, MI, USA

S. C. Wang

Division of Acute Care Surgery, Section of General Surgery, Department of Surgery, University of Michigan Health System, Ann Arbor, MI, USA

Introduction

Adrenocortical carcinoma (ACC) is a rare and deadly disease with an annual incidence of ~2 per million pop- ulation. Stage- and treatment-dependent 5-year survival rates are reported to be 13-82 % [1]. A recent study revealed that in the past 20 years no significant progress has been made with regard to the treatment of ACC, and 5-year survival outcomes have remained static [2]. Com- pleteness of resection and use of adjuvant mitotane and external beam radiotherapy improve survival in selected populations, but indications for use are still not well defined or uniformly applied [3, 4]. Accurate prediction of survival is difficult and current staging models are unreli- able. Previous oncologic studies have shown that prog- nostication regarding survival is accurate only 20 % of the

time, with most errors being overly optimistic [5, 6]. Accurate prognostication has implications related to utility of continued treatment, ability to tolerate treatment, quality of life, and timeliness of referral for hospice care at the end of life. Various ACC staging systems have been used at various times, which makes comparing the ACC literature difficult. In 2010, our group demonstrated the significant influence of tumor grade on survival [1]. Present staging systems are imprecise with regard to predicting survival as data employed in these systems are general and do not account for data on a more individualized level. Because ACC is rare, limited data exist to study other possible prognostic factors that may aid in predicting tumor recur- rence, tolerance of treatment, and overall survival, such as patient frailty, which has been found to be useful in other types of malignancy.

Sarcopenia, originally coined by Rosenberg and derived from the Greek sarx (flesh) and penia (loss), is the concept of declining function with age, the most obvious being the loss of muscle mass and strength. The concept has been classically used to examine functionality and prediction of survival in the elderly population. Central sarcopenia (CS) is a marker of overall frailty and predicts mortality. Sar- copenic findings have been independently associated with decreased function, increased falls and fractures, need for long-term care, increased health-care costs, and increased mortality [7, 8]. Sarcopenia is predictive of poor outcomes after stroke and hip fracture [8, 9]. More recently, the concept of sarcopenia has been investigated in the general adult population to determine the usefulness of muscle mass and density in predicting severity of illness, frailty, and outcomes from specific disease processes as well as tolerance for interventions such as major surgery. Previous work from the University of Michigan Analytical Mor- phomics Group in the liver transplant and abdominal aortic aneurysm populations found that prediction of outcome appeared stronger using CS markers than using other markers [10, 11]. Applied to the cancer patient, sarcopenia and frailty are intertwined with the concept of cachexia [12]. The lumbar skeletal muscle index (LSMI) has been used as a numerical metric of sarcopenia in cachectic patients.

Continuing and expanding upon our previous work involving sarcopenia, we have evaluated CS and abdominal fat distribution as it relates to the greater concept of cancer cachexia. Our hypothesis is that significant differences in psoas muscle density (PMD), psoas muscle area (PMA), lean psoas muscle area (LPMA), LSMI, intra-abdominal (IA), and subcutaneous (SC) fat exist at different stages in ACC patients. Additionally, we hypothesize that changes in PMD, PMA, LSMI, IA, and SC fat can be followed over time and correlate with disease course. Finally, we hypothesize that a model can be generated using the

existing staging system and additional morphometric variables to more accurately predict survival.

Materials and methods

This retrospective review included all patients diagnosed with stage I-IV ACC who were evaluated and treated in the University of Michigan Multidisciplinary Endocrine Oncology Clinic between July 2005 and August 2011. Demographic, treatment, imaging, laboratory, and outcome data were reviewed. Patients were divided by ENSAT stage into three groups for analysis. Stage I and II patients were combined together due to the small number of stage I patients, and previous studies have also suggested that stage I and II patients can be combined for survival anal- ysis due to the lack of discriminatory ability of the staging systems [1, 13]. Morphometric measures, including PMD [Hounsfield unit (HU)], PMA (mm2), LPMA (mm2), LSMI (cm2/m2), IA (mm), and SC (mm) fat values, were recorded from CT scans. Fifty-nine (47 %) patients had multiple scans available for review during the follow-up period. Time between scans, disease status, and changes in value of morphometric measures were analyzed to assess ability to identify change and correlate with disease state. For this study, one scan from each year of follow-up per patient was used to track morphometric changes.

Morphometric characteristics identified on CT images of the abdomen and pelvis were evaluated using highly pre- cise and cost-efficient computer-based algorithms devel- oped by UM. Hundreds of thousands of granular data elements that are one, two, and three dimensional in nature are able to be individually analyzed. Cross-sectional areas and densities of the left and right psoas muscles at the level of the fourth lumbar vertebra (L4) were processed in our study population by a single trained observer who was blinded to disease status. Outlines of the psoas muscles were traced and uploaded for image analysis. The resulting areas, densities, and other measurements were then com- puted and uploaded into a database for review. The mean L4 PMD was calculated as (left L4 mean + right L4 mean)/2. The lean PMD score was calculated using the formula (mean L4 density + 85)/170. The lean PMA (LPMA) is represented by total psoas area x lean PMD. Previous work has revealed that an adrenal mass on one side does not affect PMD or PMA on the same side com- pared to the contralateral side (p = 0.14-0.99 for all comparisons) [14]. LSMI was calculated by dividing PMA by body surface area (BSA). In addition to PMD, PMA, LPMA, and LSMI, SC fat measurements were made from the anterior abdominal wall fascia to the skin, and IA fat was measured from the L4 anterior vertebral body to the anterior abdominal wall fascia on each CT scan.

Calculations were completed using algorithms pro- grammed in MATLAB v13.0 (MathWorks, Natick, MA).

Statistical methods

Patient characteristics were summarized using means and frequencies. Pearson correlations between morphometric measures were used to examine the association between measures and to aid in highlighting potential multicollin- earity issues for the overall survival model build. A repe- ated-measures model for each morphometric measure was built to determine the relationship between the morpho- metric measure and patient age, gender, stage at diagnosis, and time of measurement in relation to date of diagnosis. This repeated-measures model included a random intercept with an unstructured covariance matrix to account for the correlation of measures within a patient. Estimates of the morphometric measure by stage are provided from the model with the corresponding type 3 p value.

Overall survival was plotted using the Kaplan-Meier method by stage. The CT scan closest to the date of diagnosis within 1 year of diagnosis was identified for each patient. Kaplan-Meier plots were made using the mor- phometric measures closest to diagnosis, dividing the population into two groups by the median value.

Cox proportional hazards models with time-dependent covariates were used to model overall survival with repe- ated morphometric measures. Univariate models were ini- tially made to find potentially important covariates. A backward model build was performed to determine the best multivariable overall survival model. Covariates examined in the model build included age, gender, stage at diagnosis, tumor grade, PMD, LPMA, LSMI, and IA fat. The mea- sures that were correlated (PMD, LPMA, and LSMI) were examined one at a time. The covariates were tested for an interaction with time. The best model included stage at diagnosis, LPMA, IA fat, and IA fat interaction with time.

All statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC). Alpha < 0.05 was used to determine statistical significance.

Results

From July 2005 to August 2011, 152 patients with ACC (94 females) were treated in our multidisciplinary endo- crine oncology clinic. One hundred twenty-five patients underwent 269 digital CT scans compatible for use with the image-processing system. The number of scans per patient ranged from 1 to 11. ACC stage at diagnosis were stage I, 2 %;stage II, 41 %;stage III, 32 %; and stage IV, 25 %. Mean age at time of diagnosis of ACC was 45.7 years (SD = 13.3, range = 18-72) and median

length of follow-up was 2.0 years (range = 1 month- 11 years) (Table 1). 68 % of the patients with stage I-III ACC who underwent surgery with curative intent had recurrence of their disease. The median length of time to first recurrence in those patients was 10.7 months (ran- ge = 0.4-59 months). Only the stage at diagnosis was significant in predicting time to recurrence. The type of surgical resection was not analyzed in this study but has previously been shown by our group to be significant with respect to recurrence in this same patient cohort. Median length of time to recurrence by stage was as follows: stage I, 32.6 months (n = 2, range = 18.9-46.3 months); stage II, 13.4 months (n = 32, range = 0.4-59 months); and stage III, 5.6 months (n = 30, range = 1.67-33.0 months). Median length of survival in years from time of diagnosis by stage was as follows: stages I and II, 10.0 [95 % CI = 4.5, 10.5]; stage III, 2.5 [1.4, 5.5]; and stage IV, 1.0 [0.5, 3.0] (Fig. 1).

Values and significance of mean PMD, LPMA, LSMI, IA, and SC fat of the cohort at different stages of disease are given in Table 2. Changes in the mean values of LPMA and IA fat between stages were found to be most signifi- cant. CS, defined as LSMI <55 cm2/m2 in men and <39 cm2/m2 in women, was investigated for 101/125 patients who had height and weight data available at the time the CT scan was performed. CS was identified at some

Table 1 Demographics for 125 patients with ACC
N = 125ACC patients
Mean age (SD)45.7 (13.3)
Gender (%)
Female74 (59.2)
Male51 (40.8)
Tumor stage (%)
I3 (2.2)
II51 (41.8)
III40 (32.0)
IV31 (24.8)
Tumor side (%)
Left71 (56.8)
Right52 (41.6)
Bilateral2 (1.6)
Received mitotane88 (70.4)
Received XRT31 (24.8)
Number of CT scans per patient (%)
166 (52.8)
232 (25.6)
310 (8.0)
45 (4.0)
5 or more12 (9.6)
Median follow-up (min, max)2.0 years (1 month-11 years)
Fig. 1 Probability of survival according to stage at diagnosis. Group 1 = Stage I and II patients, Group 2 = Stage III patients, Group 3 = Stage IV patients. 1, 2 and 5 year survival for Stage I/II patients is 96, 90, 62 % respectively. 1, 2 and 5 year survival for Stage III patients is 88, 58, and 33 %. 1, 2 and 5 year survival for Stage IV patients is 46, 40, and 16 %

1.0

0.8

Survival Probability

0.6

0.4

0.2

0.0

+ Censored

1

54

48

40

27

20

14

12

2

40

29

16

7

5

5

2

3

31

8

6

2

1

1

1

0

1

2

3

4

5

6

Years

Stage_at_Diagnosis

1: I/II

2: III

3: IV

point during the disease process in at least 54 % of patients. This may be falsely low since not all scans of all patients were able to be evaluated. At the time of diagnosis, 26.6 % of stage I and II patients had initial CT scans consistent with CS by the LSMI criterion. At the time of diagnosis, 16 % of stage III and 46 % of stage IV patients were sarcopenic. Sarcopenia did not necessarily continue or progress, and some patients who died of disease were not sarcopenic by LSMI but did show decreases in LSMI over time. Ultimately, changes in LSMI between stages was not found to be significant (p = 0.3196) (Table 2) and did not provide additional information to stage for pre- diction of length of survival. Calculation of the Pearson correlation coefficient for LSMI with LPMA revealed a highly positive correlation (r = 0.92), but was less so for PMD (r = 0.36), and almost none for IA fat (r = 0.03) and SC fat (r = 0.04). These findings are likely due to the fact that LSMI is calculated using height and weight as part of BSA as well as lumbar skeletal muscle area. Lean psoas area is calculated using psoas muscle area and a modifi- cation of the psoas density. As area is likely the most important component of the two equations, LPMA and LSMI show high correlation, but because density is not included in LSMI and is a surrogate marker for fat depo- sition that correlates with IA fat changes, correlation between LSMI and IA or subcutaneous fat (SC) is poor.

In addition to stage at diagnosis as a significant predictor of survival (p = 0.0076), morphometric variables, which on univariate analysis significantly improve the survival model compared to using stage alone, include LPMA (p = 0.0024) (Fig. 2) and IA fat (p<0.0001) (Fig. 3).

These are compiled in Table 3. The Cox regression model, including stage, LPMA, and IA fat, shows type III p values as <0.05 for each predictor, which statistically indicates that the addition of that a covariate is associated with overall survival, even after controlling for the inclusion of the other two covariates. Including IA fat by time since diagnosis is significant (p<0.0001), as is PMD (p<0.0001). An incremental 5-HU decrease in PMD suggests a 28 % increase in risk of death (hazard ratio [HR] = 0.72, 95 % CI = 0.58, 0.88). PMD is correlated to LPMA. Since both predictors cannot be in the model and other groups have found LPMA to be important and it is easier to interpret, we chose a final model using LPMA instead of PMD. We found that there is an 8 % lower hazard of death over time with an increase in LPMA of 100 mm2. This hazard of death is constant over time. Interestingly, we found that the hazard over time due to IA fat changes with time since diagnosis. At 1 year there is an additional 8 % greater hazard of death for every 2-mm increase in IA fat, at 2 years there is 3 % higher hazard of death per 2-mm increase, and at 5 years there is a 9 % lower hazard of death. This “protection” with time may be a reflection of differences in individual tumor biology, and future work may find that rate of change in morphometric variables rather than absolute value impacts length of survival and accounts for these changes in hazard of death over time.

Change in PMD, LPMA, IA, and SC fat values recorded for those patients having multiple scans and the state of disease were compared to the previous scan and revealed the ability to track changes in morphometric measures according to changes in disease status. Interval changes

Table 2 CT scan morphometric covariates by stage
Stage I/II (N = 54)Stage III (N = 38)Stage IV (N = 33)p Value*
Mean PMD (HU)56.0 (54.3, 57.6)53.3 (51.0, 55.5)50.9 (48.0, 53.7)0.006
LPMA (mm2)1680.6 (1589.8, 1771.5)1648.0 (1527.3, 1768.8)1359.2 (1210.3, 1508.2)0.001
IA fat (mm2)86.7 (80.8, 92.6)99.6 (91.9, 107.3)113.5 (103.5, 123.5)<0.001
SC (mm2)26.7 (22.7, 30.7)28.0 (23.4, 32.5)19.5 (14.3, 24.7)0.007
LSMI (cm2/m2)49.4 (45.4, 53.4)47.1 (42.3, 51.8)44.0 (38.3, 49.7)0.3196

Values reported as median (95 % CI)

HU hounsfield units

* Adjusted for age, gender, and scan time from diagnosis

between scans were identified and correlated with disease state as is shown in Table 4. The plot of patient 1 reveals significantly decreasing LPMA and increasing IA fat prior to time 0 (time of diagnosis) with stage IV ACC and deceased within 1 month. Patient 2 had severe hypercor- tisolism. His plot revealed significant improvement in LPMA and rapidly decreasing IA fat over 6 months, likely due to resolution of his hypercortisolism. There had been relative stabilization of LPMA and IA fat since that time, and accordingly he had shown no evidence of recurrence. Patient 3 had stage II disease. Although he had shown no evidence of disease recurrence since surgery and he had not reached sarcopenic levels by LSMI, his IA fat had been steadily increased and LPMA had been decreasing and he had lost 20 lb. in 3 months, indicating possible impending disease recurrence. The clinical course of patient 4 had been one of very slowly progressing stage IV disease for 10 years. His LSMI over the past few years had decreased, and the rate of change was increasing. His survival may be threatened at this point.

Discussion

There has been a recent surge in interest in the clinical application of analytical morphomics and evaluation for sarcopenia beyond that involving the aging population. The concept of sarcopenia has been investigated in cancer patients, but other novel applications are also being developed. Data generated from this study can be used to formulate models for future investigation to hopefully more accurately predict survival on an individualized basis, understand the potential course of the disease, and direct future investigations into treatment allocation for patients with ACC. Identifying the degree of frailty of the patient using analytical morphomics and understanding the trend of the disease may allow patients a better quality of life in the time they have remaining. The data presented in this study show that decreases in LPMA, PMD, and SC fat and an increase in IA fat correlate with decreased length of

survival in ACC patients. This decrease in lean muscle mass and increase in IA fat is consistent with the sarco- penic obesity identified in other types of cancer patients. Combining components of current staging systems with LPMA and IA fat may enhance the ability to more accu- rately predict length of survival in ACC patients. Surpris- ingly, LSMI was not found to add to the predictive model for survival. Disease course, including remission, stability, and progression, can be observed and plotted longitudi- nally. Careful observation of trends may also allow early prediction of a change in disease course. While this study has identified IA fat and LPMA as two measures that significantly improve the ability to predict survival, highly accurate future models will need to take into account inconsistencies in the timing of CT images, rate of change of significant variables as opposed to absolute values, effects of hormone production on morphometric measures, changes in body composition in the last year of life, and other as-yet unidentified variables.

While we have reported significant findings in this study that may have important implications for the future, we must consider this work in the context of its limitations. This was a retrospective study that used a small cohort of patients from a single institution. The effect of excess adrenal hormones on morphometric measures could not be evaluated in this study. In 2011, our group studied sarco- penia in patients with hypercortisolism and found that mean PMD and IA fat were significantly affected by 24-h urine cortisol levels in the same way that cachexia patients are affected [14]. Testosterone has been shown to increase muscle mass and improve sarcopenia. In this study we were unable to accurately assess hormone secretion as most patients were referred from elsewhere, and the vast majority did not have what we would consider the mini- mum hormonal evaluation prior to tumor resection. As many ACCs can secrete several hormones in excess, the data would be misrepresented if we used only the values of hormones that happened to have been drawn, and we were unaware of other abnormal hormone levels. One must remember that the benefit of morphomic analysis is the

Fig. 2 Overall survival by LPMA. The median value is used as the cut-off point to compare Group 1 (above the median) and Group 2 (below the median)

1.0

0.8

L

Survival Probability

0.6

0.4

0.2

0.0

+ Censored

1

47

28

21

10

5

3

2

2

47

26

14

2

1

1

1

0

1

2

3

4

5

6

Years

leanpsoasarea

1: Above Median

2: Below Median

Fig. 3 Overall survival by quantity of IA fat (mm). The median value is used as the cut- off point to compare Group 1 (above the median) and Group 2 (below the median)

1.0

0.8

Survival Probability

0.6

0.4

0.2

0.0

+ Censored

1

34

19

10

5

3

2

1

2

34

20

15

4

2

1

1

0

1

2

3

4

5

6

Years

VB2FASCIA_L4

1: Above Median

2: Below Median

effect of all comorbidities that the patient has contributes to the CT findings and a true assessment of the patient can be made. We also suspect that the rate of change over time of certain morphometric characteristics may be more impor- tant in predicting the disease course than the absolute value alone, especially since patients who are not sarcopenic or cachectic die of their disease.

Similar to the question of hormonal effects, it is unknown what the effects of mitotane or various

chemotherapy regimens used in ACC have on morphomic measures. Many patients never achieve therapeutic mito- tane levels, may not tolerate chemotherapeutic regimens, or fail chemotherapy shortly after starting, making it difficult to assess the effect of chemotherapy with respect to sar- copenia. The relationship between lean body mass as assessed by CT and patient tolerance of chemotherapy has been studied in colon cancer patients receiving 5-FU. Variations in lean body mass obtained by CT scan

Table 3 Overall survival as affected by stage, lean psoas area, IA fat, and change in fat over time using Cox regression analysis
Hazard ratio (95 % CI)p Value
Stage at diagnosis0.0076
I/II versus IV0.10 (0.04, 0.25)
III versus IV0.34 (0.15, 0.76)
LPMA (per 100 mm2 increase)0.92 (0.85, 0.99)0.0257
IA fat<0.0001
IA fata (time since diagnosis)<0.0001
IA fat at 1 year1.08 (1.04, 1.12)
IA fat at 2 years1.03 (1.01, 1.07)
IA fat at 5 years0.91 (0.87, 0.96)

a Adjusted for age, gender, and scan time from diagnosis

predicted variations in development of toxicity to 5-FU, thereby allowing dose adjustment and limiting toxicity from treatment [15].

Cachexia of cancer is defined as a multifactorial three- stage syndrome characterized by an ongoing loss of skel- etal muscle mass (with or without the loss of fat) that cannot be fully reversed by conventional nutritional sup- port and leads to progressive functional impairment [7]. The diagnosis of cachexia is supported by weight loss >5 % over 6 months, body mass index (BMI) <20, any degree of weight loss >2 %, an appendicular skeletal muscle index obtained by dual-energy X-ray absorption- etry (DEXA) consistent with sarcopenia, and any degree of weight loss >2 %. LSMI determined by CT imaging is defined as sarcopenic if measurements are <55 cm2/m2 in men and <39 cm2/m2 in women. Although bioelectrical impedance analysis and DEXA have been used historically to quantify fat and muscle mass, CT has emerged as a more precise and accessible modality to quantify body compo- sition and assess tumor changes [16, 17]. Cancer patients

are already routinely evaluated by high-resolution imaging such as CT for the purpose of diagnosis and follow-up, making morphomic analysis reasonable. Objective mea- sures of body composition, including use of muscle area and muscle density in the form of HUs as well as fat in various distributions, allow for recognition of small but potentially clinically important changes. Intramyocellular lipid droplets have been found to increase with cancer progression [18]. A decrease of 1 HU represents an increase in muscular fat deposition of 1 g/100 ml [19]. CT-based quantification of sarcopenia has already allowed identification of links between sarcopenia and functional status, chemotherapy toxicity, prediction of time to tumor progression, and mortality [20, 21].

Interestingly, evidence from this study suggests that LSMI does not perform as well as other morphometric variables for predicting survival in ACC patients. Use of LSMI will need to be further assessed in patients with ACC and other cancers. This is not entirely surprising as BSA and BMI are poor indicators of actual body composition and frailty. Additionally, a future study will prospectively test the ability to predict outcome using the identified multivariable model incorporating stage, LPMA, IA fat, and IA fat interaction with time against the traditional ACC staging system alone.

Conclusion

Survival as predicted by traditional staging systems in ACC can be more precise when considering the individual patient. The described morphometric measures correlate with results obtained using traditional ACC staging sys- tems and may add strength when used in combination. Serial measurements of morphometric data are possible in

Table 4 Disease course over time based on LPMA, IA fat measurement, and lean skeletal muscle index from four representative patients with ACC

All patients except patient 1 underwent surgical resection at the time of diagnosis (0)

PatientTime from diagnosis (years)ªLean psoas area (mm2)IA fat (mm)Lean skeletal muscle index (cm2/m2)
1-7.61240.999.243.9
0766.9150.026.5
0.1336.8128.812.1
201693.2120.049.4
0.82150.464.767.3
2.02211.664.467.5
31.32810.1111.274.0
2.12337.4102.766.3
3.32371.0102.167.0
46.62530.599.173.9
8.51793.872.355.6
9.41401.42102.045.5

a Negative number of years indicates a CT scan available for review prior to time of diagnosis

ACC patients and may allow for more individualized pre- diction of outcome. The rate of change of these variables over time may be more important than the absolute value.

Conflict of interest The authors have no conflicts of interest to disclose.

References

1. Miller BS, Gauger PG, Hammer GD et al (2010) Proposal for modification of the ENSAT staging system for adrenocortical carcinoma using tumor grade. Langenbecks Arch Surg 395: 955-961

2. Bilimoria KY, Shen WT, Elaraj D et al (2008) Adrenocortical carcinoma in the United States: treatment utilization and prog- nostic factors. Cancer 113:3130-3136

3. Sabolch A, Feng M, Griffith K et al (2011) Adjuvant and defin- itive radiotherapy for adrenocortical carcinoma. Int J Radiat Oncol Biol Phys 80:1477-1484

4. Polat B, Fassnacht M, Pfreundner L et al (2009) Radiotherapy in adrenocortical carcinoma. Cancer 115:2816-2823

5. Steensma DP, Loprinzi CL (2000) The art and science of prog- nosis in patients with advanced cancer. Eur J Cancer 36: 2025-2027

6. Christakis NA, Lamont EB (2000) Extent and determinants of error in doctors’ prognoses in terminally ill patients: prospective cohort study. Br Med J 320:469-472

7. Lang T, Streeper T, Cawthon P et al (2010) Sarcopenia: etiology, clinical consequences, intervention, and assessment. Osteoporos Int 21:543-559

8. Tosteson AN, Gottlieb DJ, Radley DC et al (2007) Excess mor- tality following hip fracture: the role of underlying health status. Osteoporos Int 18:1463-1472

9. Longstreth WT Jr, Bernick C, Fitzpatrick A et al (2001) Fre- quency and predictors of stroke death in 5,888 participants in the Cardiovascular Health Study. Neurology 56:368-375

10. Englesbe MJ, Patel SP, He K et al (2010) Sarcopenia and mor- tality after liver transplantation. J Am Coll Surg 211:271-278

11. Lee JS, He K, Harbaugh CM et al (2011) Frailty, core muscle size, and mortality in patients undergoing open abdominal aortic aneurysm repair. J Vasc Surg 53:912-917

12. Fearon K, Strasser F, Anker SD et al (2011) Definition and classification of cancer cachexia: an international consensus. Lancet Oncol 12:489-495

13. Lughezzani G, Sun M, Perrotte P et al (2010) The European Network for the Study of Adernal Tumors is prognostically superior to the international union against cancer-staging system: a North American validation. Eur J Cancer 46:713-719

14. Miller BS, Ignatoski KM, Daignault S et al (2011) A quantitative tool to objectively assess degree of sarcopenia in patients with hypercortisolism. Surgery 150(6):1178-1185

15. Prado CM, Lieffers JR, McCargar LJ et al (2008) Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a popula- tion-based study. Lancet Oncol 9:629-635

16. Mourtzakis M, Prado CM, Lieffers JR et al (2008) 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 33:997-1006

17. Prado CM, Birdsell LA, Baracos VE (2009) The emerging role of CT in assessing cancer cachexia. Curr Opin Support Palliat Care 3:269-275

18. Stephens NA, Skipworth RJ, Macdonald AJ et al (2011) Intra- myocellular lipid droplets increase with progression of cachexia in cancer patients. J Cachexia Sarcopenia Muscle 2:111-117

19. Goodpaster BH, Kelley DE, Thaete FL et al (2000) Skeletal muscle attenuation determined by computed tomography is associated with skeletal muscle lipid content. J Appl Physiol 89:104-110

20. Prado CM, Baracos VE, McCargar LJ et al (2007) Body com- position as an independent determinant of 5-fluorouracil-based chemotherapy toxicity. Clin Cancer Res 13:3264-3268

21. Andreyev HJ, Norman AR, Oates J et al (1998) Why do patients with weight loss have a worse outcome when undergoing che- motherapy for gastrointestinal malignancies? Eur J Cancer 34:503-509