Psoas muscle variables and adjuvant chemotherapy-related and survival outcomes following pancreaticoduodenectomy for periampullary malignancy

Abstract

BACKGROUND

Early skeletal muscle wasting has been demonstrated to be associated with chemotherapy-related toxicity in patients with cancer. Hounsfield unit average calculation (HUAC) or psoas muscle index (PMI) can be used to diagnose skeletal muscle wasting radiologically by calculations involving psoas muscle area.

AIM

To evaluate ability of HUAC and PMI in predicting adjuvant chemotherapy (AC)-related and survival outcomes following pancreaticoduodenectomy for periampullary malignancy.

METHODS

A retrospective study of patients who underwent pancreaticoduodenectomy for periampullary malignancy was conducted. HUAC and PMI were determined from preoperative computed tomography scans. Time to AC, completion of AC, survival, failure to initiate AC, mortality during AC, AC-related toxicity, cancer progression during AC and non-chemotherapy-related complications were the outcome measures.

RESULTS

A total of 148 patients were included of whom 62.2% completed AC. The AC was terminated due to toxicity in 33.9%, cancer progression in 25.0%, non-chemotherapy-related complications in 16.1%, or poor performance status in 3.5% of the patients. The 1-,3-,5-years survival rates were 85.1%, 29.7% and 19.9%, respectively. The overall median survival was 22.6 (76.8) months. HUAC was an independent predictor of time to AC (P = 0.0005), AC-related toxicity (P = 0.0431), and completion of AC (P = 0.0486). However, HUAC did not predict 1-year (P = 0.8616), 3-years (P = 0.5941) or overall (P = 0.9206) survival. Being an octogenarian (P = 0.0009), Clavien-Dindo > III (P = 0.0184), and length of hospital stay (P = 0.0103) were independent predictors of failure to initiate AC. Moreover, being an octogenarian was an independent predictor of 3-years (P = 0.0201) and overall (P = 0.0433) survival. PMI did not predict any of the outcome measures.

CONCLUSION

Unlike PMI, preoperative HUAC, determined from psoas muscle area and density, may predict time to AC, AC-related toxicity and completion of AC following pancreaticoduodenectomy for periampullary cancers. However, its ability to predict survival should be evaluated by future research.

Key Words: Pancreaticoduodenectomy; Adjuvant chemotherapy; Hounsfield unit average calculation; Psoas muscle index; Periampullary cancer

Core Tip: Preoperative Hounsfield unit average calculation, determined from psoas muscle area and density, may predict time to adjuvant chemotherapy (AC), AC-related toxicity and completion of AC following pancreaticoduodenectomy for periampullary cancers. However, its ability to predict survival should be evaluated by future research. This study aims to evaluate ability of Hounsfield unit average calculation and psoas muscle index in predicting AC-related and survival outcomes following pancreaticoduodenectomy for periampullary malignancy.

INTRODUCTION

Pancreatoduodenectomy is a technically complex procedure which is carried out for both benign and malignant disease of the pancreatic head and periampullary region[1]. Considering that pancreaticoduodenectomy is associated with high risk of postoperative morbidities, several studies have been conducted to identify predicting factors that can improve short-term and long-term outcomes of pancreaticoduodenectomy[2,3].

Following pancreaticoduodenectomy for periampullary cancers, adjuvant chemotherapy (AC) is recommended to optimize overall patient’s survival although confirmation of such survival benefits for early-stage cancer still requires high quality stage specific research[4-6]. Although the optimum time to initiate AC following pancreaticoduodenectomy should be within 12 weeks of surgery, delay in the initiation of AC beyond 12-weeks has not been demonstrated to negatively impact the survival[7,8]. However, termination of AC before its completion or lower number of AC cycles have been demonstrated to be associated with lower survival in resected pancreatic cancer[7,9].

Sarcopenia, aging-related loss of skeletal muscle mass, has been recognised as an independent predictor of postoperative morbidity and mortality in emergency or elective abdominal surgery, including pancreaticoduodenectomy for periampullary malignancies[10]. Furthermore, even early skeletal muscle wasting in the absence of diagnosis of sarcopenia has been demonstrated to be associated with chemotherapy-related toxicity in patients with pancreatic cancer[11].

Psoas muscle index (PMI) or Hounsfield unit average calculation (HUAC) can be used to diagnose sarcopenia radiologically by calculations involving psoas muscle area[12]. Radiologically-measured psoas muscle mass variables have been able to predict cardiopulmonary exercise test (CPET) performance and long-survival in colorectal cancer patients. Nevertheless, such predictive abilities have not been comprehensively researched in periampullary cancer patients. Furthermore, their abilities to predict outcomes in patients with resected periampullary cancer with satisfactory preoperative CPET performance is an interesting subject to explore. The aim of this study was to investigate whether HUAC or PMI can predict AC-related and survival outcomes after pancreaticoduodenectomy in patients with periampullary malignancies.

MATERIALS AND METHODS

Reporting standards

The study was conducted in line with standards recommended by the Strength the Reporting of Observational Studies in Epidemiology statement[13].

Study design

A retrospective analysis of prospectively collected data was conducted for consecutive patients who underwent pancreaticoduodenectomy for periampullary malignancy at our tertiary center from January 2017 to January 2023. Considering that the study had a retrospective design involving non-identifiable data from hospital databases, approval by Research Ethics Committees and patient consent were not needed. This study was compliant with the Helsinki medical research ethical principles and policies recommended by the local Clinical Governance Unit.

Population and condition of interest

Patients of any age or gender who had AC following pancreaticoduodenectomy for periampullary malignancy were considered for inclusion. Any histologically confirmed periampullary malignancy including pancreatic ductal adenocarcinoma, distal cholangiocarcinoma, ampullary adenocarcinoma or duodenal adenocarcinoma were considered. Patients who had pancreaticoduodenectomy for benign or pre-malignant pathologies were excluded. Moreover, the patients that had other types of cancers including neuroendocrine tumours, gastrointestinal stromal tumours, or metastatic lesions to periampullary regions from other primary cancers were excluded considering the different adjuvant strategies that such patients receive. All the included patients underwent CPET before the surgery and considered for surgery if they fulfilled peak VO₂ of at least 14 mL/minute/kg and anaerobic threshold of at least 10 mL/minute/kg.

Prognostic factors of interest

HUAC which was determined by two independent radiologists using Sectra PACS IDS7 software from the most recent (within 2 months of surgery) preoperative post-contrast computed tomography scan in the portal venous phase for each patient at the level of L3 using the following equation: (Right psoas area × density) + (left psoas area × density)/total area. PMI which was determined by two independent radiologists from the most recent (within 2 months of surgery) preoperative post-contrast computed tomography scan in the portal venous phase for each patient by dividing total psoas muscle area at the level of L3 by height of the patient. The psoas muscle area was determined using Sectra PACS IDS7 software.

Outcomes

The primary outcome parameters were time to adjuvant AC defined as time from the date of surgery to the date of initiation of AC, successful completion of AC, and survival (1-year, 3-years, 5 years, and overall). Failure to initiate AC for any reasons other than patient’s choice, mortality during AC, AC-related toxicity leading to termination of treatment, cancer progression during AC causing termination of AC, and none-chemotherapy related complications leading to termination of AC were the evaluated secondary outcome measures.

Collection of data

The patients’ related data were collected by two authors independently, and a third author was consulted in case of disagreement. The collected data included: Baseline characteristic of the study populations including age, gender, body mass index, the American Society of Anesthesiologists (ASA) grade, use of neoadjuvant chemotherapy, procedure, type of malignancy, stage of the cancer, and degree of tumour differentiation, presence of cardiac disease, or presence of chronic obstructive pulmonary disease (COPD). Postoperative information including overall preoperative complications as per Clavien-Dindo (C-D) classification[14], postoperative pancreatic fistula (POPF) grades A, B, or C as per definition of the International Study Group on Pancreatic Fistula[15,16], mortality (30-day, 90-day and in hospital), readmission, re-operation, length of hospital stay (LOS).

Statistical analysis

We applied simple descriptive statistics was applied to evaluate demographics, clinical characteristics and outcome data. Data regarding patient’s demographic and clinical characteristics were summarized with means ± SD for continuous variables and n (%) for categorical variables. Multivariable logistic regression model (binary regression when dependent variable was dichotomous and linear regression when dependent variable was continuous) was constructed in which the primary outcome measures were defined as dependent variables and the following variables were defined as independent variables: HUAC, PMI, sarcopenia, age, being an octogenarian (age ≥ 80), ASA status, clinically relevant (CR)-POPF, C-D > III, cardiac disease, COPD and LOS. Statistical significance was assumed at P < 0.05. Data were analysed using MedCalc Statistical Software version 19.2.6 (MedCalc Software bv, Ostend, Belgium).

RESULTS

Baseline characteristics

A total of 259 patients underwent pancreaticoduodenectomy for periampullary cancer. Of those, 148 patients (57.4%) received AC. Table 1 presents the baseline characteristics of the study population. The mean age of the included patients was 68.3 ± 8.29 years. There were 81 (55.0%) male and 67(45.0%) female patients. The ASA grade was I in 7 (4.7%), II in 62 (42.0%), III in 77 (52.0%), and IV in 2 (1.3%) patients. Only 2 (1.3%) patients received neoadjuvant chemotherapy. The indication for pancreaticoduodenectomy was pancreatic cancer in 94 (63.5%), distal cholangiocarcinoma in 15 (10.1%), duodenal cancer in 9 (6.1%), and ampullary cancer in 30 (20.3%) patients. The stage of cancer was I in 23 (15.6%), II in 116 (78.3%), and III in 9 (6.1) patients. The tumours were poorly-differentiated in 81 (54.7%), moderately-differentiated in 59 (39.9%), well-differentiated in 5 (3.3%), or undifferentiated in 2 (1.4%) patients.

Table 1 Baseline characteristics of the included patients, mean ± SD/n (%).

Characteristic	Total (n = 148)
Age	68.3 ± 8.29
Gender (male/female)	81 (55)/67 (45)
ASA
1	7 (4.7)
2	62 (42.0)
3	77 (52.0)
4	2 (1.3)
BMI (kg/m2)	26.2 ± 4.0
Neoadjuvant chemotherapy	2 (1.3)
Procedure
PPPD	137 (92.5)
Classical Whipple	11 (7.5)
Type of malignancy
Pancreatic	94 (63.5)
Cholangiocarcinoma	15 (10.1)
Duodenal	9 (6.1)
Ampullary	30 (20.3)
TNM stage
I	23 (15.6)
II	116 (78.3)
III	9 (6.1)
IV	0 (0)
Poor	81 (54.7)
Moderate	59 (39.9)
Well	5 (3.3)
Undifferentiated	2 (1.4)
Unable to assess	1 (0.7)
R status
0	68 (45.9)
1	79 (53.4)
2	1 (0.7)
Number of harvested lymph nodes	20 ± 8.3

ASA: American Society of Anesthesiologists; BMI: Body mass index; PPPD: Pylorus preserving pancreaticoduodenectomy; TNM: Tumor, node, metastasis.

Postoperative information

Postoperative information is presented by Table 2. Overall perioperative complications: A total of 75 (50.1%) postoperative complications happened. There were 25 (16.9%) C-D I, 41 (27.7%) C-D II, and 8 (5.4%) C-D III, and 1 (0.7%) C-D IV complications. POPF: Of 33 (22.3%) patients developed POPF. There were 21 (14.2%) grade A, 12 (8.1%) grade B, and no grade C POPF events. The rate of CR-POPF was 8.1%. Postoperative haemorrhage: The postoperative haemorrhage happened in 3 (2.0%) patients. Bile leak: There was no bile leak in this study. Chyle leak: There was no chyle leak in this study. Mortality: There was no 30-days, 90-days, or in-hospital mortalities in this study. Re-operation: None of the included patients required a re-operation or return to theatre. LOS: The mean LOS of the included patients was 14 ± 6.2 days. Re-admission to hospital: Eighteen patients (12.2%) required re-admission to hospital.

Table 2 Postoperative outcomes, n (%).

Characteristic	Total (n = 148)
Postoperative complications
C-D I	25 (16.9)
C-D II	41 (27.7)
C-D III	8 (5.4)
C-D IV	1 (0.7)
POPF
A	21 (14.2)
B	12 (8.1)
C	0 (0)
Chyle leak	0 (0)
Bile leak	0 (0)
Postoperative hemorrhage	3 (2.0)
Mortality
30-day	0 (0)
90-day	0 (0)
In-hospital	0 (0)
LOS, mean ± SD (days)	14 ± 6.2
Re-operation	0 (0)
Re-admission	18 (12.2)
Survival
1-yaer	126 (85.1)
3-years	44 (29.7)
5-years	19 (19.8)

C-D: Clavien-Dindo; POPF: Postoperative pancreatic fistula; LOS: Length of hospital stay.

Prognostic factors

PMI: The median PMI was 559.5 (1009.8). The mean PMI was 576.5 ± 153.7. HUAC: The median HUAC was 43.03 (65.2) days. The mean HUAC was 41.9 ± 10.0.

Outcome synthesis

The results of outcome syntheses are presented in Tables 3, 4 and 5.

Table 3 Multivariable logistic regression analysis of independent variables for prediction of time to adjuvant chemotherapy, completion of adjuvant chemotherapy, 1-year, 3-years and overall survival.

Variable	Time to AC		Completion of AC		1-year survival		3-years survival		Overall survival
	Standardized coefficient	P value	Standardized coefficient	P value	Standardized coefficient	P value	Standardized coefficient	P value	Standardized coefficient	P value
PMI	-0.0109	0.2405	-0.0003	0.1985	0.0001	0.3790	0.0001	0.9774	0.3140	0.3065
HUAC	0.5074	0.0005	0.0124	0.0486	0.0005	0.8616	0.0020	0.5941	0.4704	0.9206
Sarcopenia	6.3857	0.1676	0.2114	0.1188	-0.0059	0.9531	0.0551	0.6641	-33.5165	0.8262
Age	0.1439	0.4434	-0.0036	0.5082	-0.0074	0.0689	0.0010	0.8328	-1.6205	0.7927
Octogenarian	2.9289	0.7158	0.2488	0.2923	0.2749	0.1151	0.5160	0.0201	538.2032	0.0433
ASA	1.7999	0.4676	0.0005	0.9941	0.0031	0.9535	0.0326	0.6292	45.9027	0.5729
CR-POPF	7.2686	0.1739	0.1050	0.5017	0.0878	0.4456	0.0525	0.7182	228.5744	0.1931
C-D > III	5.9580	0.3445	0.0764	0.6788	-0.0841	0.5361	-0.0448	0.7943	-106.1970	0.6078
LOS	0.6556	0.0076	-0.0033	0.6392	-0.0010	0.8462	-0.0101	0.1290	-7.6509	0.3385
Cardiac disease	2.8105	0.5780	-0.1286	0.3855	-0.0572	0.5997	-0.2473	0.0747	-291.7777	0.0804
COPD	-9.9204	0.1640	0.4382	0.0368	0.1614	0.2935	0.1030	0.5954	111.0465	0.6344

AC: Adjuvant chemotherapy; PMI: Psoas muscle index; HUAC: Hounsfield unit average calculation; ASA: American Society of Anesthesiologists; CR-POPF: Clinically relevant-postoperative pancreatic fistula; C-D: Clavien-Dindo; LOS: Length of stay; COPD: Chronic obstructive pulmonary disease.

Table 4 Multivariable logistic regression analysis of independent variables for prediction of failure to initiate adjuvant chemotherapy following pancreaticoduodenectomy.

Variable	Failure to initiate AC
	Standardized coefficient	P value
PMI	0.0003	0.1041
HUAC	-0.0014	0.6104
Age	-0.0001	0.9826
Octogenarian	-0.4071	0.0009
ASA	-0.0668	0.1881
CR-POPF	0.0612	0.5631
C-D > III	-0.2514	0.0184
LOS	-0.0083	0.0103
Cardiac disease	0.1013	0.3200
COPD	-0.2022	0.0842

AC: Adjuvant chemotherapy; PMI: Psoas muscle index; HUAC: Hounsfield unit average calculation; ASA: American Society of Anesthesiologists; CR-POPF: Clinically relevant-postoperative pancreatic fistula; C-D: Clavien-Dindo; LOS: Length of stay; COPD: Chronic obstructive pulmonary disease.

Table 5 Multivariable logistic regression analysis of independent variables for prediction of failure to complete adjuvant chemotherapy due to mortality toxicity, disease progression, or non-chemotherapy-related complications.

Variable	Mortality		Toxicity		Disease progression		NCRC
	Standardized coefficient	P value	Standardized coefficient	P value	Standardized coefficient	P value	Standardized coefficient	P value
PMI	0.0002	0.0651	-0.0001	0.7255	-0.0004	0.7784	0.0002	0.1798
HUAC	-0.0013	0.4073	-0.0059	0.0431	0.0025	0.8818	0.0003	0.8697
Sarcopenia	0.0173	0.7343	-0.0284	0.7622	-0.1182	0.1523	-0.0213	0.7528
Age	0.0006	0.7556	0.0028	0.4599	-0.1658	0.4587	0.0006	0.8234
Octogenarian	-0.0085	0.9228	0.0238	0.8839	-0.0107	0.2495	-0.0770	0.5109
ASA	-0.0165	0.5436	-0.0316	0.5292	-0.1152	0.8092	0.0471	0.1925
CR-POPF	0.0279	0.6328	-0.0923	0.3929	-0.1012	0.2268	0.0568	0.4641
C-D > III	0.0974	0.1593	0.1013	0.4269	0.0051	0.3683	-0.1046	0.2545
LOS	-0.0025	0.3498	0.0043	0.3783	0.0931	0.2442	-0.0035	0.3274
Cardiac disease	0.0469	0.3976	-0.0308	0.7635	-0.0004	0.3024	-0.0833	0.2578
COPD	-0.0491	0.5281	-0.1133	0.4311	0.0025	0.7784	-0.0484	0.6396

NCRC: Non-chemotherapy-related complications; PMI: Psoas muscle index; HUAC: Hounsfield unit average calculation; ASA: American Society of Anesthesiologists; CR-POPF: Clinically relevant-postoperative pancreatic fistula; C-D: Clavien-Dindo; LOS: Length of stay; COPD: Chronic obstructive pulmonary disease.

Primary outcomes

Time to AC. The median time to AC was 76 (144) days. The mean time to AC was 75.7 ± 17.9 days. Completion of chemotherapy: A total of 92 (62.2%) patients completed AC. The AC was not completed due to mortality in 5 (8.9%), non-chemotherapy related complications in 9 (16.1%), disease progression in 14 (25.0%), poor performance status in 2 (3.5%), side effects in 9 (16.1%), or toxicity in 19 (33.9%) patients. Overall survival: The overall median survival was 22.6 (76.8) months. The mean survival was 28.0 ± 17.5 months. 1-, 3-, 5-year survival: The 1-, 3-, 5 years survival rates were 85.1%, 29.7% and 19.9%, respectively.

Predictors of primary outcomes

Time to AC: Multivariable linear logistic regression identified HUAC (standardized coefficient: 0.5074, P = 0.0005) and LOS (standardized coefficient: 0.6556, P = 0.0076) as independent predictors of time to AC. However, PMI (standardized coefficient: -0.0109, P = 0.2405), sarcopenia (standardized coefficient: 6.3857, P = 0.1676), age (standardized coefficient: 0.1439, P = 0.4434), being an octogenarian (standardized coefficient: 2.9289, P = 0.7158), ASA grade (standardized coefficient: 1.7999, P = 0.4676), CR-POPF (standardized coefficient: 7.2686, P = 0.1739), C-D > III (standardized coefficient: 5.9580, P = 0.3445), cardiac disease (standardized coefficient: 2.8105, P = 0.5780), or COPD (standardized coefficient: -9.9204,P = 0.1640) did not predict time to AC.

Completion of AC: Multivariable binary logistic regression identified HUAC (standardized coefficient: 0.0124, P = 0.0486) or COPD (standardized coefficient: 0.4382, P = 0.0368) as independent predictors of completion of AC. However, PMI (standardized coefficient: -0.0003, P = 0.1985), sarcopenia (standardized coefficient: 0.2114, P = 0.1188), age (standardized coefficient: -0.0036, P = 0.5082), being an octogenarian (standardized coefficient: 0.2488, P = 0.2923), ASA grade (standardized coefficient: 0.0005, P = 0.9941), CR-POPF (standardized coefficient: 0.1050, P = 0.5017), C-D > III (standardized coefficient: 0.0764, P = 0.6788), LOS (standardized coefficient: -0.0033, P = 0.6788), or cardiac disease (standardized coefficient: -0.1286, P = 0.3855) did not predict completion of AC.

One-year survival: Multivariable binary logistic regression demonstrated that HUAC (standardized coefficient: 0.0005, P = 0.8616), PMI (standardized coefficient: 0.0001, P = 0.3790), sarcopenia (standardized coefficient: -0.0059, P = 0.9531), age (standardized coefficient: -0.0074, P = 0.0689), being an octogenarian (standardized coefficients: 0.2749, P = 0.1151), ASA grade (standardized coefficient: 0.0031, P = 0.9535), CR-POPF (standardized coefficient: 0.0878, P = 0.4456), C-D > III (standardized coefficient: -0.0841, P = 0.5361), LOS (standardized coefficient: -0.0010, P = 0.8462), cardiac disease (standardized coefficient: -0.0572, P = 0.5997), or COPD (standardized coefficient: 0.1614, P = 0.2935) did not predict 1-year survival.

Three-years survival: Multivariable binary logistic regression identified being an octogenarian (standardized coefficient: 0.5160, P = 0.0201) as an independent predictor of 2-years survival. However, HUAC (standardized coefficient: 0.0020, P = 0.5941), PMI (standardized coefficient: 0.0001, P = 0.9774), sarcopenia (standardized coefficient: 0.0551, P = 0.6641), age (standardized coefficient: 0.0010, P = 0.8328), ASA grade (standardized coefficient: 0.0326, P = 0.6292), CR-POPF (standardized coefficient: 0.0525, P = 0.7182), C-D > III (standardized coefficient: -0.0448, P = 0.7943), LOS (standardized coefficient: -0.0101, P = 0.1290), cardiac disease (standardized coefficient: -0.2473, P = 0.0747), or COPD (standardized coefficient: 0.1030, P = 0.5954) did not predict 3-years survival.

Overall survival: Multivariable linear logistic regression identified being an octogenarian (standardized coefficient: 538.2032, P = 0.0433) as an independent predictor of overall survival. However, HUAC (standardized coefficient: 0.4704, P = 0.9206), PMI (standardized coefficient: 0.3140, P = 0.3065), sarcopenia (standardized coefficient: -33.5165, P = 0.8262), age (standardized coefficient: -1.6205, P = 0.7927), ASA grade (standardized coefficient: 45.9027, P = 0.5729), CR-POPF (standardized coefficient: 228.5744, P = 0.1931), C-D > III (standardized coefficient: -106.1970, P = 0.6078), LOS (standardized coefficient: -7.6509, P = 0.3385), cardiac disease (standardized coefficient: -291.7777, P = 0.0804), or COPD (standardized coefficient: 111.0465, P = 0.6344) did not predict overall survival.

Secondary outcomes

Failure to initiate AC: A total of 111 (42.8%) patients did not initiate AC. Mortality during AC: The AC was not completed due to mortality in 5(8.9%) patients. AC-related toxicity: A total of 19 (33.9%) patients developed AC related toxicity leading to termination of AC. Progression of cancer during AC: 14 (25.0%) patients had progression of cancer during AC which led to termination of AC. None-chemotherapy related complications: Non-chemotherapy related complications happened in 9 (16.1%) patients leading to the termination of AC. Poor performance status: 2 (3.5%) patients did not complete AC due to poor performance status.

Predictors of secondary outcomes

Failure to initiate AC: Multivariable binary logistic regression identified being an octogenarian (standardized coefficient: -0.4071, P = 0.0009), C-D > III (standardized coefficient: -0.2514, P = 0.0184), and LOS (standardized coefficient: -0.0083, P = 0.0103), as independent predictor of failure to initiate AC. However, HUAC (standardized coefficient: -0.0014, P = 0.6104), PMI (standardized coefficient: 0.0003, P = 0.1041), age (standardized coefficient: -0.0001, P = 0.9826), ASA grade (standardized coefficient: -0.0668, P = 0.1881), CR-POPF (standardized coefficient: 0.0612, P = 0.5631), cardiac disease (standardized coefficient: 0.1013, P = 0.3200), or COPD (standardized coefficient: -0.2022, P = 0.0842) did not predict failure to initiate AC.

Mortality during AC: Multivariable binary logistic regression demonstrated that HUAC (standardized coefficient: -0.0013,P = 0.4073), PMI (standardized coefficient: 0.0002, P = 0.0651), sarcopenia (standardized coefficient: 0.0173, P = 0.7343), age (standardized coefficient: 0.0006, P = 0.7556), being an octogenarian (standardized coefficient: -0.0085, P = 0.9228), ASA grade (standardized coefficient: -0.0165, P = 0.5436), CR-POPF (standardized coefficient: 0.0279, P = 0.6328), C-D > III (standardized coefficient: 0.0974, P = 0.1593), LOS (standardized coefficient: -0.0025, P = 0.3498), cardiac disease (standardized coefficient: 0.0469, P = 0.3976), or COPD (standardized coefficient: -0.0491, P = 0.5281) did not predict mortality during AC.

AC related toxicity: Multivariable binary logistic regression identified HUAC (standardized coefficient: -0.0059, P = 0.0431) as independent predictors of AC related toxicity as cause of termination of AC. However, PMI (standardized coefficient: -0.0001, P = 0.7255), sarcopenia (standardized coefficient: -0.0284, P = 0.7622), age (standardized coefficient: 0.0028, P = 0.4599), being an octogenarian (standardized coefficient: 0.0238, P = 0.8839), ASA grade (standardized coefficient: -0.0316, P = 0.5292), CR-POPF (standardized coefficient: -0.0923, P = 0.3929), C-D > III (standardized coefficient: 0.1013, P = 0.4269), LOS (standardized coefficient: 0.0043, P = 0.3783), cardiac disease (standardized coefficient: -0.0308, P = 0.7635), or COPD (standardized coefficient: -0.1133, P = 0.4311) did not predict AC related toxicity as cause of termination of AC.

Progression of cancer causing termination of AC: Multivariable binary logistic regression demonstrated that HUAC (standardized coefficient: 0.0025, P = 0.8818), PMI (standardized coefficient: -0.0004, P = 0.7784), sarcopenia (standardized coefficient: -0.1182, P = 0.1523), age (standardized coefficient: -0.1658, P = 0.4587), being an octogenarian (standardized coefficient: -0.0107, P = 0.2495), ASA grade (standardized coefficient: -0.1152, P = 0.8092), CR-POPF (standardized coefficient: -0.1012, P = 0.2268), C-D > III (standardized coefficient: 0.0051, P = 0.3683), LOS (standardized coefficient: 0.0931, P = 0.2442), cardiac disease (standardized coefficient: -0.0004, P = 0.3024), or COPD (standardized coefficient: 0.0025, P = 0.7784) did not predict disease progression as cause of termination of AC.

None-chemotherapy related complications: Multivariable binary logistic regression demonstrated that HUAC (standardized coefficient: 0.0003, P = 0.8697), PMI (standardized coefficient: 0.0002, P = 0.1798), sarcopenia (standardized coefficient: -0.0213, P = 0.7528), age (standardized coefficient: 0.0006, P = 0.8234), being an octogenarian (standardized coefficient: -0.0770, P = 0.5109), ASA grade (standardized coefficient: 0.0471, P = 0.1925), CR-POPF (standardized coefficient: 0.0568, P = 0.4641), C-D > III (standardized coefficient: -0.1046, P = 0.2545), LOS (standardized coefficient: -0.0035,P = 0.3274), cardiac disease (standardized coefficient: -0.0833, P = 0.2578), or COPD (standardized coefficient: -0.0484,P = 0.6396) did not predict none-chemotherapy related complications as cause of termination of AC.

DISCUSSION

In view of growing evidence in favor of the ability of radiologically-measured psoas muscle mass variables in predicting the outcomes of patients undergoing surgical oncological resections, we conducted a retrospective study to investigate ability of preoperative HUAC or PMI in predicting AC-related and survival outcomes in patients undergoing pancreaticoduodenectomy for periampullary malignancies at our center over a 5-years period. The subsequent analyses demonstrated that preoperative HUAC was an independent predictor of time to initiation of AC, completion of AC and AC-related toxicity leading to termination of treatment before its completion. However, HUAC was not able to predict patient’s survival. Furthermore, our findings indicated that preoperative PMI did not predict time to AC, completion of AC, or survival outcomes.

The conducted multivariable regression analyses identified being an octogenarian, C-D complications > III and LOS as independent predictors of failure to initiate AC. Moreover, being an octogenarian was found to be an independent predictor of 3-years and overall survival following pancreaticoduodenectomy for periampullary cancers. However, we did not find age, CR-POPF, cardiac disease or COPD as predictors of any of the outcome measures.

Both PMI and HUAC are used for radiological diagnosis of sarcopenia which not only has been demonstrated to be a negative predictor of adverse outcomes in abdominal surgeries, including pancreaticoduodenectomy, but also has been considered in recent risk predication models for emergency abdominal surgeries[10,11,17]. However, although PMI and HUAC are accurate in categorical determination of sarcopenia, their predictive abilities in patients without sarcopenia seem to be much more limited following pancreaticoduodenectomy. In this study, all patients had satisfactory CPET test and their median PMI and HUAC were 559.5 and 43.03, respectively, which were above the defined cut-off values for diagnosis of sarcopenia indicating that most of the patients who underwent pancreaticoduodenectomy for periampullary cancers were not sarcopenic. Nevertheless, a patient may lose significant amount of skeletal muscle mass without diagnostic values of PMI or HUAC for sarcopenia. This was the rationale behind using psoas muscle mass variables rather than sarcopenia as the main prognostic factors of interest in this study. Aberle et al[11] evaluated the relationship between body composition and chemotherapy-related toxicity in advanced stage pancreatic cancer patients and demonstrated that early skeletal muscle mass loss was associated with treatment-related toxicity. In agreement with these findings, we found that HUAC was able to predict AC-related toxicity causing termination of treatment which happened in 33.9% of the patients. This may also explain the demonstrated ability of HUAC in predicting completion of AC in this study. Although we have demonstrated that HUAC can predict time to initiation of AC, we do not believe this is of important clinical significance as delay in the initiation of AC has not been shown to negatively impact the survival following pancreaticoduodenectomy[7,8].

AC is known to improve survival following pancreaticoduodenectomy for periampullary cancer[7,9]. Only 57.4% of our patients who underwent pancreaticoduodenectomy for periampullary malignancies received AC. Although the preoperative HUAC or PMI did not predict failure to initiate AC, we identified being an octogenarian, C-D complications > III and LOS as independent predictors of failure to initiate AC. These findings are consistent with the existing literature. A recent meta-analysis of best available evidence demonstrated that pancreaticoduodenectomy in octogenarians has been shown to be associated with higher mortality rate, postoperative complications, and reduced likelihood to undergo adjuvant therapy[18].

The findings of this study suggest that in patients with satisfactory CPET performance who undergo pancreaticoduodenectomy for periampullary cancer, HUAC has a potential to predict AC-related outcomes including AC-related toxicity and, most importantly, completion of such therapy. Although we have not demonstrated predictive ability of HUAC regarding survival outcomes, future high-quality research in this context is needed. We encourage future studies to evaluate the predictive ability of postoperative HUAC rather than the preoperative value prior to initiation of AC as the postoperative HUAC values may represent more relevant skeletal muscle mass before the initiation of the AC. In fact, the preoperative HUAC may be affected by several confounding factors including patient’s response to surgery, postoperative complications, nutritional status, etc., which can alter the skeletal muscle mass that is determined preoperatively. Furthermore, we encourage future studies to evaluate the predictive ability of HUAC on outcomes of neoadjuvant chemoradiotherapy in patients with pancreatic cancer.

The current study has some limitations that should be considered when interpreting its findings. It was a retrospective study with its associated inherited selection bias. The sample size of our study was relatively small which subject our findings to bias. Although the calculations of HUAC and PMI were performed by two independent radiologists, human error might have introduced some bias to the calculated values. The study was a single-center study and the generalizability of findings should be done with caution due to single-center nature of the study. Furthermore, we were not able to conduct conducted multivariable regression analysis with respect to 5-year survival as such data for some of the patients were not yet determined considering our study period. Moreover, we did not report our outcome with respect to tumor size, tumor morphology, complexity of surgery, etc., which can impact survival. Finally, we have included all types of periampullary cancers which might have introduced bias when evaluating survival outcomes considering their different potentials.

CONCLUSION

The findings of this retrospective study suggest that preoperative HUAC value, determined from psoas muscle area and density, may predict time to AC, AC-related toxicity and completion of AC following pancreaticoduodenectomy for periampullary cancers. However, its ability to predict survival in such patients is in doubt. However, preoperative PMI does not seem to predict AC-related or survival outcomes. Future studies should concentrate on impact of pre-AC rather than preoperative HUAC values to eliminate bias associated with the confounding factors, including postoperative complications or patient’s nutritional status, that can negatively impact skeletal muscle mass from the time of pancreaticoduodenectomy to AC.