Relationships Among Body Mass Index, Longitudinal Body Composition Alterations, and Survival in Patients With Locally Advanced Pancreatic Cancer Receiving Chemoradiation. A Pilot Study

Abstract

Context

In pancreatic cancer, the presence of obesity or weight loss is associated with higher mortality.

Objectives

To explore the relationships among body mass index (BMI), longitudinal body composition alterations, and clinical outcomes in pancreatic cancer patients.

Methods

Records of 41 patients with inoperable, locally advanced pancreatic cancer who participated in a prospective chemo-radiation study were reviewed. Body composition was analyzed from two sets of computed-tomography images obtained before and after radiation treatment (median interval 104 days).

Results

Median age was 59 years, and 56% of patients female. Twenty-four (59%) patients were either overweight (22%) or obese (37%). Sarcopenia was present in 26 (63%) patients. At follow-up, weight loss was experienced by 33 (81%) patients. The median losses (%) before and after treatment were: weight 5% (P< 0.001), skeletal muscle (SKM) 4% (P=0.003), visceral adipose tissue (VAT) 13% (P< 0.001), and subcutaneous adipose tissue (SCAT) 11% (P=0.002). SKM loss positively correlated with age (P=0.03), baseline BMI (P < 0.001), and VAT (P=0.04) index. Obese patients experienced higher losses in weight (P=0.009), SKM (P=0.02), and VAT (P=0.02). Median survival was 12 months. In univariate analysis, age, baseline obesity, sarcopenic obesity, and losses (%) in weight, SKM, and VAT were associated with worse survival. In multivariate analysis, only age (hazard ratio (HR)=1.033, P=0.04 and higher VAT loss (HR=2.6, P=0.03) remained significant.

Conclusion

Our preliminary findings suggest that obese patients experience higher losses in weight, SKM and VAT, which may contribute to poorer survival in these patients.

Introduction

Cancer cachexia is a complex metabolic disorder characterized by involuntary weight loss from skeletal muscle (SKM) and adipose tissue (AT) depletion, anorexia, fatigue, anemia, and impairments in immune, endocrine and physical functioning.^1,2 A majority of inoperable pancreatic cancer patients already experience significant weight loss by the time of cancer diagnosis,³ which progresses through the trajectory of illness,⁴ and has been associated with increased morbidity and mortality.^5–8 Paradoxically, having a high body weight or body mass index (BMI) does not appear to be protective. There is accumulating evidence that obesity is a risk factor for developing pancreatic cancer and cancer-related mortality.^9–11

At the time of pancreatic cancer diagnosis, obese patients, particularly those with visceral adiposity, may already be predisposed to metabolic derangements and higher morbidity/mortality as in the non-cancer setting. High visceral AT (VAT) is positively associated with inflammation, metabolic derangements, muscle loss and increased cardiovascular/thrombotic morbidity, which has been attributed to its higher expression and release of adipokines and free fatty acids (FFA) into the portal circulation. ^12–19 Recent studies employing computed tomography (CT) images for body composition analysis in pancreatic ^20,21 and other cancers^22,23 suggest that overweight/obese patients with sarcopenia,^20,22,23 or patients with high VAT, ²¹ have poorer survival. VAT also is recognized to be most responsive to lipolytic factors.^16,24,25 The presence of cancer-induced lipolytic factors ²⁶ in viscerally obese cancer patients theoretically could result in high amounts of VAT lipolysis, release of FFA and other mediators, thereby exacerbating inflammation, metabolic derangements, and muscle wasting. Increased VAT loss has been observed in weight- losing cancer patients. ^27,28 Accordingly, we hypothesized that pancreatic cancer patients with high BMI would have higher VAT and SKM losses, inflammation, metabolic/endocrine derangements, and poorer survival.

To test our hypothesis, we have first conducted this pilot study to examine the relationship between BMI and the losses in SKM and AT, with clinical outcomes of symptoms and survival, in a homogeneous cohort of pancreatic cancer patients who received similar cancer therapy. A better understanding of the association between body composition alterations and clinical outcomes across a range of body weights and BMI is warranted in pancreatic cancer, as these may offer prognostic information and insights into cachexia mechanisms, thereby allowing formulation of more effective treatment strategies.

Patients and Methods

This study was reviewed and approved by the Institutional Review Board of the University of Texas, M. D. Anderson Cancer Center.

Patient Sample and Study Design

We analyzed data from a prospective cohort of patients with inoperable, locally advanced pancreatic cancer, who were participants in a previously reported Phase I chemo-radiation trial,²⁹ and a conjoined prospective symptom assessment study.³⁰ The Phase I trial investigated the safety/tolerability of bevacizumab in combination with capecitabine and radiation.²⁹ Bevacizumab was administered two weeks before radiotherapy, every two weeks during radiotherapy, and continued until disease progression. Radiotherapy was administered at a dose of 50.4 Gy in 28 fractions delivered to the primary tumor and peri-pancreatic adenopathy over five and a half weeks. Capecitabine was administered with radiotherapy on Days 14–52. As reported previously,²⁹ eligibility criteria for study participants included a Karnofsky Performance Status (KPS) score of 70 or greater, and 46 of 48 (96%) patients had baseline KPS of 90 or greater. Forty-three patients consented to the parallel symptom assessment study, and as reported, ³⁰ patients who did not consent for the symptom study did not differ from the study population for demographics or disease stage. Of these 43 patients, 41 had evaluable CT scans, in the defined period of the study, and constituted our study sample.

Anthropometric and Body Composition Measurements

Weight and height were recorded according to standard methods. BMI was calculated as weight (in kilograms, kg) divided by height (in meters, m) squared (kg/m²). Standard international cutoffs for BMI ³¹ were used to classify patients into healthy weight or normal (BMI 18.5–24.9), underweight (BMI less than 18.5), overweight (BMI 25 to 29.9 kg/m²), and obese (BMI 30 or more).

Body composition was analyzed from two sets of CT scans, obtained prior to initiation of chemo-radiation, and following completion of radiation therapy, respectively. Four consecutive images obtained at the third lumbar vertebra (L-3) and inferiorly, as done previously, ³² were used to quantify and average for tissue cross-sectional areas, corresponding to SKM, VAT and subcutaneous adipose tissue (SCAT) depots. L-3 has been used as a landmark in prior studies of body composition analysis in cancer.^20,23,32,33 Tissue types were automatically segmented by using CT thresholds (Hounsfield units) of −29 to 150 for SKM, −50 to 150 for VAT, and −190 to −30 for SCAT. Tissue boundaries were then reviewed and manually corrected as necessary. Normalized tissue cross-sectional area (cm²) was computed by summing the given tissue’s pixels and multiplying the sum by the absolute unit pixel surface area. SKM and AT cross-sectional areas also were normalized for stature, and reported as the respective tissue index (cm²/m²).

The baseline SKM index was used to categorize patients as sarcopenic, based on previously defined cutoff L-3 SKM index values (less than 38.5 cm²/m² for females, and less than 52.4 cm²/m² for males). ^20,23 All weight and tissue changes are reported as the percentage change per 100 days. The percentage change in tissue index per 100 days was calculated by dividing the percentage change in tissue index by the time interval between the first and second CT images, and multiplying this number by 100.

Symptom Assessment

Symptoms were assessed prospectively using the M. D. Anderson Symptom Inventory (MDASI) as part of the conjoined symptom assessment study.³⁰ Fatigue and anorexia clustered together during the course of the study, suggesting a strong interrelatedness between these two symptoms.³⁰ As anorexia and fatigue are common in cancer cachexia, these two symptoms were selected for our analysis from two sets of MDASI assessments, which were performed around the first day of chemotherapy (median 0 days, interquartile range [IQR] −1 to 7 days) and a median of 49 days (IQR 43 to 52 days) later. The MDASI has been validated, and is a brief measure of the self-reported severity and impact of cancer-related symptoms.^34,35 Each symptom is rated on an 11-point scale, with 0 being absence of symptom, and 10 being the “worst score possible.”

Laboratory Data

Serum albumin and hemoglobin levels obtained a median of 0 days (IQR 0 to 16) prior to the first CT scan, and 0 days (IQR 0 to 15) prior to the second CT scan, were used in the analysis.

Statistical Analysis

Descriptive statistics were used to summarize patients’ demographic, anthropometric, body composition, laboratory, and symptom parameters. Comparison between patient groups was assessed using Kruskal-Wallis nonparametric testing for continuous variables and the Chi-squared test for categorical variables. Spearman’s rank correlation coefficients (two-tailed test of significance) were used to quantify associations among variables. Overall survival in months was calculated from the dates of first and second CT scans to the date of death or last known to be alive. Survival from the date of the first CT scan was used for survival analysis. The Kaplan-Meier logrank test was used to compare the survival between subgroups. The relationship between predictive variables and survival was assessed using multivariate Cox regression analysis. Backward stepwise model was used for variable selection, and a cutoff value of 0.05 was used for inclusion in the model. A value of P < 0.05 was considered statistically significant. The Statistical Package for the Social Sciences (SPSS version 17, SPSS, Inc., Chicago, IL) was used for statistical analysis.

Results

Selected baseline and follow-up characteristics of 41 patients are presented in Table 1. The median age was 59 years (range 42–81 years). The majority (95%) of patients were Caucasian, and 56% were female. The median baseline BMI was 25.8 kg/m² (IQR 23.2–29.7). Based on BMI values, 17 (42%) patients were in the normal range, 15 (37%) were overweight, and nine (22%) obese. There were no patients in the underweight category (BMI less than 18.5 kg/m²). Sarcopenia was present in 26 (63%) patients overall: 11 (65%) patients who were of normal weight, nine (60%) who were overweight and six (67%) who were obese (P=0.94)

Table 1.

Baseline and Post-Treatment Patient Characteristics (n=41)

	All patients (n=41)
Age in years median (range)	58.9 (41.7–81.0)
Male/female, n	18/23
Ethnicity, n Caucasian/Other	39/3
Weight in kg, median (IQR)	75.7 (61.8–85.3)
BMI in kg/m2, median (IQR)	25.8 (23.2–29.7)
BMI category, n (%)
- BMI 18.5 to 24.9	17 (41%)
- BMI 25 to 29.9 kg/m2	15 (37%)
- BMI ≥30	8 (20%)
L3 tissue area in cm2, median (IQR)
- SKM	122.3 (100.7 145.3)
- VAT	126.7 (92.5–205.0)
- SCAT	167.3 (91.6–244.1)
Calculated tissue index in cm2/m2, median (IQR)
- SKM index	43.6-38.0-47.1)
- VAT index	46.0 (33.2–69.2)
- SCAT index	55.3 (31.9–81.5)
Sarcopenic, n (%)	15 (37)
Baseline symptoms
- Fatigue	2 (1–4)
- Anorexia	2 (0–5)
Baseline albumin g/dL	4.0 (3.7–4.2)
Adjusted weight/tissue losses over 100 days, median (IQR)
- % weight loss	4.7 (0.3–10.0)
- % SKM loss	3.8 (−0.9–10.1)
- % VAT loss	12.9(1.3–31.5)
- % SCAT loss	11.1 (−1.9–25.3)

BMI = body mass index; IQR = interquartile range; SCAT = subcutaneous adipose tissue; SKM = skeletal muscle; TAT = total adipose tissue; VAT = visceral adipose tissue.

The median interval between the first and second CT scan was 104 days (IQR 97–112). At follow-up, weight loss (more than 0) was experienced by 33 (81%) patients. The median percentage loss in body weight was 4.7% (IQR 0.3–10.0; P< 0.001), and accompanied by significant losses in SKM (median 3.8%, IQR −0.9 to 10.1; P=0.003), VAT (12.9%, IQR 1.3 to 31.5; P< 0.001), and SCAT (11.1%, IQR −1.9 to 25.3; P=0.002) indices (Table 1). Figure 1 shows the distribution of the percentage changes in SKM and adipose tissue indices for our study cohort. Fourteen (34%) patients had stable/gain in SKM. Nine (22%) and 11 (27%) patients had stable/gain in VAT and SCAT, respectively. As shown in Table 2, percent of SKM loss positively correlated with age (P=0.03), BMI (P < 0.001), and VAT (P=0.04) index, and the losses in weight (P=0.014), VAT and SCAT indices (P < 0.001). Percentage of weight loss correlated positively with the losses (%) in SCAT (P<0.001) and SKM (P=0.02), but not with VAT loss (P=0.10). There was also a small but significant increase in median fatigue score (P=0.012) at follow-up, and a decrease in median serum albumin levels (P=0.015) (data not shown).

Fig. 1.

Distribution of the percentage changes in skeletal and adipose tissue per 100 days for patients with locally advanced pancreatic cancer (n=41).

Table 2.

Association Between Baseline BMI, Age, and the Losses in Skeletal Muscle, Adipose Tissue, and Weight

Variables	% SKM Loss		% VAT loss		% SAT loss		% Weight loss
	r	P	r	P	r	P	r	P
Age	0.35	0.03	0.00	1.00	0.33	0.04	0.24	0.12
Baseline BMI	0.53	<0.001	0.38	0.01	0.32	0.04	0.24	0.13
Baseline SKM index	0.18	0.26	0.06	0.70	−.07	0.66	0.04	0.80
Baseline VAT index	0.32	0.04	0.58	<0.001	0.09	0.58	0.12	0.48
Baseline SCAT index	0.26	0.10	0.06	0.73	0.49	<0.001	0.28	0.08
% Weight loss	0.38	0.02	0.26	0.10	0.54	<0.001	-	-
% SKM loss	-	-	0.52	<0.001	0.53	<0.001	0.38	0.02
% VAT loss	0.52	<0.001	-	-	0.15	0.37	0.26	0.10
% SCAT loss	0.53	<0.001	0.15	0.37	-	-	0.54	<0.001

Spearman’s correlation coefficient (significance two-tailed)

BMI = body mass index; SCAT = subcutaneous adipose tissue; SKM = skeletal muscle; VAT = visceral adipose tissue.

Patient Characteristics According to Baseline BMI Status

As shown in Table 3, obese patients had higher median baseline VAT (P<0.001) and SCAT (P=0.014) indices, and experienced significantly higher losses in weight (median 10% versus 4%, P=0.009), SKM (10% versus 2 %, P=0.02), and VAT (31% versus 11%, P=0.02).

Table 3.

Patient Characteristics by BMI Category

Variable	BMI <30 kg/m2	BMI ≥30 kg/m2	P-value
• % SKM loss	1.5 (−3.1–7.9)	10.4 (3.5–13.9)	0.020
• % VAT loss	10.6 (−7.8–25.5) n=32	31.3 (12.8–51.7) n=9	0.022
• % SCAT loss Age, yrs, median (range)	9.6 (−2.6–20.2) 57.7(41.7–81.0)	15.6(8.6–30.8) 64.8 (43.6–77.5)	0.174 0.816
Male/female, n/n	14/18	4/5	0.970
Weight in kg, median (IQR)	72.4 (59.0–80.3)	94.0 (82.5–99.0)	<0.001
BMI in kg/m2, median (IQR)	25 (23–27)	32 (30–32)	<0.001
Calculated tissue index in cm2/m2, median (IQR)
• SKM index	42.4 (36.3–46.8)	44.4 (42.5–51.5)	0.128
• VAT index	42.3 (31.4–58.9)	82.1 (58.8–98.3)	<0.001
• SCAT index	46.1 (30.0–74.2)	77.3 (56.8–118.0)	0.014
Baseline symptoms
• Fatigue	2 (1–3)	3 (0–7)	0.505
• Anorexia	1 (0–4)	4 (0–6)	0.415
Baseline albumin g/dL	4.0 (3.7–4.2)	4.0 (3.8–4.3)	0.816
Sarcopenic, n (%)	12 (38)	3 (33)	0.819
Adjusted weight and tissue losses over 100 days, median (IQR)
• % Weight loss	3.6 (0–8.0)	10.3 (5.0–14.3)	0.009

BMI = body mass index; IQR = interquartile range; SCAT = subcutaneous adipose tissue; SKM = skeletal muscle; TAT = total adipose tissue; VAT = visceral adipose tissue.

Survival Analysis

Median survival from the dates of the first CT and the second CT scans were 12.0 (IQR 9.0–18.7) and 8.5 (5.6–15.9) months, respectively. On univariate analysis, higher losses in SKM (4% or more, 0.01) and VAT (≥13%, 0.04), baseline obesity (logrank 0.01), and sarcopenia in obese patients (logrank 0. 004) were associated with poorer survival, whereas sarcopenia alone was not significant (logrank 0.246) (Fig. 3 and Table 4). On multivariate analysis, only age (hazard ratio [HR]=1.03, P=0.04) and higher VAT loss (HR 2.06, P=0,03) remained significant predictors of poorer survival (Table 4). Among patients who experienced VAT loss (n=32), only seven patients had stable/gain in SKM. The median survival for patients who gained/stable SKM and those who lost SKM was 16.8 and 10.7 months, respectively, but this was not found to be significant (logrank 0.19).

Table 4.

Hazard Ratios for Risk of Death Associated With Age, Sex and Body Composition Parameters in Patients With Locally Advanced Pancreatic Cancer (n=41)

Variable	Univariate Analysisa			Multivariate Analysisb
	HR	95% CI	P	HR	95% CI	P
Age	1.031	0.999–1.063	0.055	1.033	1.001–1.066	0.044
Sex	1.288	0.677–2.449	0.441
Percentage weight loss > or < median	1.845	0.967–3.521	0.063	1.808	1.060–3.650	0.098
Percentage SKM loss > or < median	2.083	1.088–4.000	0.027
Percentage VAT loss > or < median	1.957	1.034–3.690	0.039	2.062	1.056–4.032	0.034
Percentage SCAT loss > or < median	0.851	0.453–1.603	0.618
Sarcopenic vs. non-sarcopenic	1.477	0.761–2.874	0.249
Obese vs. overweight/normal	2.681	1.201–5.988	0.016
Obese/overweight vs. normal	1.095	0.576–2.084	0.782
Sarcopenic and obese vs. others	3.717	1.421–9.708	0.004	1.808	0.937–3.650	0.054

SCAT = subcutaneous adipose tissue; SKM = skeletal muscle; TAT = total adipose tissue; VAT = visceral adipose tissue. All tissue losses are over 100 days

^aCox univariate analysis.

^bBackward conditional method of Cox proportional hazards model. Variables included in the26 multivariate analysis include age, percentage losses in weight, SKM, VAT, baseline obesity,

Discussion

Among nutritional prognostic factors, the presence of obesity^9–11 or weight loss ^5–8 has been shown to be associated with higher pancreatic cancer mortality. Involuntary weight loss in cancer patients occurs as part of a complex metabolic syndrome, characterized by progressive losses in muscle and adipose tissues, anorexia, fatigue, and impairments in immune, metabolic, endocrine and physical functioning.^1,36 Paradoxically, excess body weight or obesity does not appear to confer any protection.^9–11 Unintentional weight loss in obese cancer patients initially may go unrecognized, or even be considered to be beneficial by both patients and their providers. The main objective of our pilot study was to examine the relationship between baseline BMI and longitudinal alterations in body composition parameters, and to explore survival outcomes in a homogeneous cohort of patients with inoperable, locally advanced pancreatic cancer.

At baseline, the majority of patients (59%) were either obese (22%) or overweight (37%), and none of the patients were undernourished. The absence of underweight patients in our cohort could be explained by the selection process of the parent Phase I study. ²⁹ However, with the exponential increase in the prevalence of obesity in the U.S. in the past decades, ³⁷ and the accumulating evidence linking obesity status with increased risk for pancreatic cancer development, ^9,10,38,39 a higher number of patients with pancreatic cancer with high BMI values could be expected. Indeed, in another study, where the majority (93%) of 111 pancreatic cancer patients had stage IV disease, only 10% of patients were found to be underweight, based on BMI values.²⁰ In our study, weight loss was experienced by the majority of patients, and accompanied by significant losses in SKM and AT. Furthermore, obese patients as compared with the non-obese group, experienced disproportionately higher losses in weight (median 10% versus 4%), VAT (31% versus 11%), and SKM (10% versus 2%).

In the non-cancer setting, the link between obesity, particularly visceral obesity, and metabolic dysfunction, higher disease risk, and mortality is well established. ^{12,14,16–19,40,41} Plausible mechanisms implicating VAT include its higher responsiveness to lipolytic factors, resulting in increased release of adipokines and FFA into the portal circulation and liver, contributing to a pro-inflammatory state. ^{16,24–26,42} Systemic inflammation has been associated with various metabolic/endocrine derangements (such as insulin resistance, and hypogonadism), sarcopenia, increased cardiovascular and thrombotic morbidity, and poorer survival.^{13,15,17,19,41} Voluntary weight loss in obese non-cancer patients, through pharmacological and life-style measures, that results in VAT loss has been recognized to be beneficial, occurring in association with improvements in insulin resistance and stable/increase muscle mass.^43–47 High VAT loss has been observed in weight-losing cancer patients,^27,48 but its prognostic implication, and relationship with BMI and other body composition parameters, in cancer patients is not well understood.

Obese patients, when diagnosed with pancreatic cancer, already may have or be predisposed to metabolic derangements, muscle loss, and higher morbidity/mortality. In a recent study, the presence of high amounts of VAT on CT imaging obtained prior to surgery for pancreatic cancer was associated with increased likelihood of death.²¹ In another study conducted in patients with advanced cancer, accelerated loss in adipose tissue was shown to be predictive of poorer survival.²⁸ Increased cancer-induced lipolysis resulting from proinflammatory cytokines, neuroendocrine activation, and other tumor-induced lipolytic factors (such as zinc-α2-glycoprotein [ZAG]), appear to play an important role in adipose tissue wasting.⁴⁹ In our cohort, the losses (%) in SKM positively correlated with baseline BMI and VAT index, and the losses (%) in adipose tissues. A physiologically important cross-talk between adipose tissue and skeletal muscle has been suggested to play a role in the context of cancer cachexia.⁵⁰ It is plausible that therapies that target cancer-induced lipolytic factors could improve immune-metabolic parameters and decrease muscle loss. Indeed, in a recent preclinical study involving tumor mouse models with high cytokines and ZAG, genetic ablation of adipose triglyceride lipase prevented the increase in lipolysis and adipose tissue breakdown, and was accompanied by preservation of skeletal muscle mass.⁵¹ In our sample, a small subset of patients (n=7) experienced VAT loss that was not accompanied by SKM loss, but rather stable/gain in SKM. These patients had a trend toward better survival, which is of interest and should be further explored.

At baseline, we found sarcopenia to be present in about two-thirds of our sample, and was equally distributed among the three BMI categories. In recent years, several studies in pancreatic ²⁰ and other cancers ^22,23 have reported on sarcopenia being a frequent finding in cancer patients. In these studies, presence of sarcopenia in obese²³ or overweight/obese^20,22 patients was found to be associated with poorer survival, whereas sarcopenia or high BMI alone had no discernable association with mortality. As with these studies, we did not find sarcopenia alone to be a prognostic factor, and the presence of sarcopenia in obese patients was associated with poorer survival on univariate analysis. However, in our study, baseline obesity, and the losses in SKM and VAT, also were associated with lower survival in univariate analysis, but only higher VAT loss (and age) remained significant in multivariate analysis.

The finding of VAT loss as a prognostic factor is in agreement with another study, where losses in adipose tissue were associated with poorer survival. ²⁸ These findings, however, differ from the study by Tan et al.,²⁰ where a subset of patients with pancreatic cancer (n=44) who had longitudinal body composition analysis did not find any association with survival based on losses in adipose or muscle. It also differs from other studies where sarcopenia and obesity/overweight remained significant prognostic factors. Our unique findings may be related to the difference in the patients’ cancer therapy and in their trajectory of illness. Our cohort comprised patients with locally advanced disease, with good performance status, and who had a median survival of 12 months, whereas in the study by Tan el al., the majority (93%) of patients had metastatic disease, with overall median survival of approximately six months. The international consensus on cancer cachexia recognizes the different phases of cachexia (precachexia, cachexia, and refractory cachexia),³⁶ and it is possible body composition changes (muscle or adipose tissue) have a variable impact on survival based on different time points in the trajectory of illness. Further research is needed to better understand the curves of relative loss of muscle and adipose tissue over time. Second, all our patients received bevacizumab in combination with capecitabine as part of the parent trial. Some cancer-directed therapies, such as bevacizumab and sorafenib, have been shown to contribute to weight and muscle loss. ^52,53 In a non-randomized study of colorectal cancer patients, bevacizumab in conjunction with conventional chemotherapy was associated with loss of weight and SKM, which was independent of cancer progression.⁵² Our findings suggest that patients receiving chemotherapies combined with targeted therapies, such as bevacizumab, may experience losses in SKM muscle that are not a result of tumor cell byproducts or proinflammatory cytokines, and, therefore, these changes may not be associated with survival, as may be expected for cancer cachexia-related muscle and weight changes. Another explanation as to why sarcopenia with obesity did not remain significant as a prognostic factor might be related to the fact that sarcopenia was present in the majority of patients, and our sample size was too small to address this.

We are aware of several limitations to our study. First, the sample size was limited by the nature of the prospective Phase I study on which this analysis was based. However, our cohort consists of a homogeneous group of patients with pancreatic cancer who had inoperable, locally advanced disease and who were receiving similar treatments. Some key known prognostic factors for advanced pancreatic cancer, such as baseline KPS (96% patients had KPS of 90 or more in the original study), albumin (median 4 g/dL), and hemoglobin (median 12 g/dL), were homogeneous in our cohort, and thus were not incorporated in our multivariate analysis. Second, our retrospective analysis of CT images for body composition requires further validation. We consider our findings to be preliminary and plan to confirm our results in future research.

Fig. 2.

Kaplan-Meier survival curves according to the median percentage losses in A) SKM (≤ or >4%), B) VAT ≤ or >13%), C) SAT (≤ or >11%), and D) BMI category (BMI < or ≥30).