Pancreatic and periampullary cancers are associated with poor prognosis with 10% to 20% 5-year survival [1]. Most patients are asymptomatic in the early stages, and are often too advanced at the point of diagnosis. However, with advances in neoadjuvant therapies and extended resections, it is possible to achieve complete resection rates in up to 60% to 80% of cases. Resections, often in traditional or modified Whipple’s procedure, are technically complex, and require long anesthetic duration and longer recovery times in the hospital and at home.

A vital aspect of the preoperative assessment process is risk stratification. This process can help objectify the clinical triage process, and quantify the probability of severe morbidity and mortality. The Eastern Cooperative Oncology Group/World Health Organization Performance Status (ECOG/WHO PS) [2], although a well-established global marker of morbidity and mortality, is descriptive, variably assessed, and lacks numerical quantification [2]. Furthermore, it cannot discriminate between physically active patients and nutritionally depleted candidates. Traditionally, nutritional risk assessment is carried out using biochemical and anthropometric measures. It is well recognized that before cancer surgery, a considerable proportion of patients have a poor nutritional state. This is particularly the case in patients with pancreatic cancer, with more than 80% of patients suffering from significant weight loss at diagnosis, and developing severe cachexia over time [3]. This is likely due to a combination of the systemic effects of malignancy, the mass effect, and the complex psychosocial impact of malignancy. Myosteatosis is a measure of deposition of adipose tissue within muscle, and has been shown to decrease functional status and provide a marker of overall frailty and nutrition [4]. Myosteatosis changes can be visualized in computed tomography (CT) studies in the form of low attenuation changes within the musculature, which are considered to occur before the onset of functional decline and frailty [5]. Myosteatosis has been shown to affect short- and long-term outcomes in colorectal and oesophagogastric surgery [6,7].

While there has been a lack of published research regarding the influence of myosteatosis on long-term outcomes in those with operable pancreatic cancer, several studies have evaluated the link within the general pancreatic cancer population. In a retrospective observational study, Kim et al. [8] investigated the significance of myosteatosis and sarcopenia on 330 patients with metastatic pancreatic cancer receiving palliative chemotherapy, and observed that the overall survival was related to myosteatosis and sarcopenia. They also observed grade 3 or higher chemotherapy-induced toxicity in patients with myosteatosis (44% vs. 60%, p = 0.023). A meta-analysis from Aleixo et al. [9], which included over 21,000 patients, highlighted the significant impact of myosteatosis on survival in all cancer patients. However, less than 2% of the included patients were those with pancreatic cancer, highlighting the considerable lack of research on this topic. The impact of myosteatosis in the context of non-operative pancreaticoduodenal malignancies and their survival has been demonstrated in previous studies [8,10,11]. However, there is limited evidence of its effect on the short- and long-term outcomes in patients undergoing pancreatoduodenectomy (PD) for malignant indications. This study aims to evaluate if the presence of preoperative myosteatosis leads to poorer short- and long-term outcomes in those patients who undergo PD for suspected/confirmed malignant tumors.

This is a retrospective review of all patients who underwent PD for suspected pancreatic, ampullary, distal bile duct cholangiocarcinoma, or malignant duodenal tumors at University Hospitals Plymouth NHS Trust (UHP) between July 2006 and May 2013. The UHP Audit department approved the study as a service evaluation project, and adhered to the standards laid down in the Declaration of Helsinki (revised 2013). Patients were excluded if the indication for PD was benign or pre-malignant tumors. However, patients whose subsequent histology confirmed benign or pre-malignant tumors are included in the study. Patients were also excluded if they underwent distal pancreatectomy or total pancreatectomy as their primary procedure, or were lost to follow-up before five years. All patients underwent classic pancreaticoduodenectomy with a retro colic hepaticojejunostomy, pancreaticogastrostomy, and antecolic gastrojejunostomy. All patients were discussed in a multidisciplinary team meeting before consideration of surgery. The following data were obtained from the hepatobiliary unit’s prospectively maintained database, patients’ medical notes, and electronic records: patient demographics, comorbidities, details of preoperative imaging and staging, procedure and intraoperative findings, postoperative complications, histology, and five-year survival. This project was registered as a service evaluation project with our local department and IRB approval was not deemed necessary. This was due to all data being fully anonymised and no patient identifiable information was collected.

Postoperative pancreatic fistula (POPF) was defined as per the International Study Group of Pancreatic Surgery (ISGPS) [12]. Bile leak was defined per the International Study Group for Liver Surgery (ISGLS) [13]. Postoperative pancreatic hemorrhage (PPH) and delayed gastric emptying were defined as per the ISGPS. The gastro-jejunal leak was confirmed intraoperatively, or by contrast, extravasation on CT. Patients were considered to have a chest infection if antibiotics were prescribed during their index admission for a clinically diagnosed chest infection. The intra-abdominal collection was confirmed by either CT or ultrasound. Patients were considered to have had a surgical site infection if treated with either antibiotics, opening of the wound, or both, during the index admission.

Assessment of myosteatosis

The Picture Archiving and Communication System was utilized to retrieve the immediate preoperative CT scan. The body composition analysis was performed using OSIRIX software ver. 9.0 (OSIRIX), the widely available standard assessment tool. A single slice CT image at the level of the L3 vertebra was evaluated using Hounsfield unit (HU) thresholds of (−29 and 150) for skeletal muscle. The mean muscle attenuation in HU was reported for the whole muscle area at the L3 vertebra level. Myosteatosis was defined, as per the previous literature, as < 38.5 HU for males, and < 36.1 HU for females [14].

Statistical analysis

Differences between myosteatosis status and the patient’s baseline characteristics and comorbidity status were tested using Fisher’s exact test of proportions. Associations between myosteatosis and conditions coded as binary indicator variables were analyzed using logistic regression, adjusted for age and sex using Stata SE v.14.2 (Stata Co.), deriving the risk ratios and risk differences from the odds ratios [15]. Estimates were presented at 95% confidence intervals to aid interpretation of the effect sizes. Survival times to the endpoints of the end of surgery, the removal of a nasogastric (NG) tube, and death and discharge dates were analyzed using Cox regression. In the case of the latter two endpoints, the survival times could be censored by the end of the follow-up. The assumption of proportional hazards was tested following Grambsch and Therneau [16] to compare the slope of the scaled Schoenfeld residuals against time to zero with a chi-squared distribution.

Patient characteristics

A total of 214 patients were included in the study. The cohort’s median age was 62 years (41−80 years), similar in both groups. In toto, 120 patients (56.1%) had myosteatosis, and 94 had no evidence of myosteatosis. There were proportionally slightly more males (55.6%) in both groups. Table 1 shows the comorbidity profiles. There were significantly more patients with an ASA grade of 3 (27.5% vs. 17.0%) and fewer with grade 1 (5.0% vs. 14.9%) in the myosteatosis group (p = 0.019). The proportion of males in the myosteatosis group was also significantly smaller (45.0% vs. 69.1%; p = 0.001). A slightly greater proportion of patients with myosteatosis had been diagnosed with cardiovascular and respiratory diseases, but otherwise, there were no significant differences in baseline characteristics between the groups. There was no significant difference between the groups regarding the tumor site (Table 2) (p = 0.74). There was also no significant difference between the groups in the T or N staging (Table 3) (p = 0.30 and 0.87, respectively). The median duration of surgery was 300 minutes in both groups (range: 180−720 minutes in the myosteatosis group vs. 180−570 minutes in the non-myosteatosis group).

Table 1 . Details of patient’s characteristics and comorbidities

Patient characteristic		Non-myosteatosis	Myosteatosis	Total	p-value
Sex	Female	29 (30.9)	66 (55.0)	95 (44.4)	0.001
	Male	65 (69.1)	54 (45.0)	119 (55.6)
Preoperative stenting	No	23 (24.5)	29 (24.2)	52 (24.3)	> 0.999
	Yes	70 (74.5)	90 (75.0)	160 (74.8)
	Missing	1 (1.1)	1 (0.8)	2 (0.9)
ASA grade	1	14 (14.9)	6 (5.0)	20 (9.3)	0.019
	2	54 (57.4)	66 (55.0)	120 (56.1)
	3	16 (17.0)	33 (27.5)	49 (22.9)
	Missing	10 (10.6)	15 (12.5)	25 (11.7)
Cardiovascular disease	No	76 (80.9)	90 (75.0)	166 (77.6)	0.327
	Yes	18 (19.1)	30 (25.0)	48 (22.4)
Respiratory disease	No	86 (91.5)	102 (85.0)	188 (87.9)	0.206
	Yes	8 (8.5)	18 (15.0)	26 (12.1)
Diabete	No	87 (92.6)	114 (95.0)	201 (93.9)	0.567
	Yes	7 (7.4)	6 (5.0)	13 (6.1)

Values are presented as number (%).

Table 2 . Details of postoperative histology and comparison between the groups

	Myosteatosis	Non-myosteatosis
Benign	8 (6.7)	10 (8.3)
Pancreatic adenocarcinoma	53 (44.2)	37 (30.8)
Cholangiocarcinoma	16 (13.3)	15 (12.5)
Ampullary carcinoma	26 (21.7)	25 (20.8)
Duodenal carcinoma	2 (1.7)	1 (0.9)
Metastases	1 (0.9)	0 (0)
Periampullary (exact origin is not clear)	14 (11.7)	6 (5.0)

Values are presented as number (%).

There was no significant difference between the groups regarding the tumor site (p value = 0.74).

Table 3 . T and N stage of the primary tumors on postoperative histology

	Myosteatosis	Non-myosteatosis	p-value
T stage			0.30
T1	4 (4.9)	1 (1.7)
T2	8 (9.9)	10 (16.9)
T3	69 (85.2)	48 (81.4)
N stage			0.87
N0	28 (29.8)	20 (28.6)
N1	66 (70.2)	50 (71.4)

Values are presented as number (%).

There was no significant difference between the groups in the T or N staging.

Short-term (< 90 days) outcomes

The mortality was similar in the two groups (3.3% vs. 3.2%, p = 0.95). There was also no difference in the median overall hospital stay and intensive care unit stay (ICU). Of the whole cohort (n = 214), 90 patients required admission to the ICU, whereupon the median ICU stay was 2 days (range: 1−14 days), with no difference between the two groups. The median hospital stay for both groups was 12 days (range: 5−56 days for the myosteatosis group vs. range 5−82 days for the non-myosteatosis group). We did not observe differences between groups in the median length (4 days, range 1−30 days) of NG tube drainage.

The risk of a pancreatic leak was higher in the non-myosteatosis group (29.8% vs. 13.3%; p = 0.006). There was also a significantly greater rate of postoperative bleeding in the non-myosteatosis group (13.8% vs. 5.0%; p = 0.034). However, there was no significant difference in other postoperative complications, including intra-abdominal collection, bile leak, and chest infection; the rate of postoperative interventions (total parental nutrition, further interventions required) was also similar between the two groups (Table 4). There was a non-significant higher wound infection rate (16.0% vs. 8.3%; p = 0.114) and lower blood transfusion rate (11.7% vs. 18.5%; p = 0.180) in the non-myosteatosis group.

Table 4 . Comparison of outcomes between both groups after adjusting for age and sex

Outcome	Group		Odds ratio (95% CI)	p-value	Risk ratio (95% CI)	p-value	Risk difference (95% CI)	p-value
	Non-myosteatosis	Myosteatosis
Blood transfusion	11/94 (11.7)	22/119 (18.5)	1.735 (0.776, 3.879)	0.180	1.597 (0.801, 3.185)	0.184	0.069 (–0.029, 0.168)	0.165
Intra-abdominal collection	13/94 (13.8)	14/120 (11.7)	0.791 (0.342, 1.829)	0.583	0.816 (0.396, 1.684)	0.582	–0.026 (–0.119, 0.067)	0.586
Bile leak	5/94 (5.3)	7/120 (5.8)	1.013 (0.299, 3.439),	0.983	1.013 (0.321, 3.190)	0.983	0.001 (–0.064, 0.065)	0.983
Pancreatic leak	28/94 (29.8)	16/120 (13.3)	0.371 (0.182, 0.754)	0.006	0.458 (0.260, 0.807)	0.007	–0.160 (–0.273, –0.046)	0.006
Chest infection	6/94 (6.4)	9/120 (7.5)	1.219 (0.403, 3.687)	0.726	1.200 (0.432, 3.335)	0.727	0.013 (–0.057, 0.083)	0.724
Wound infection	15/94 (16.0)	10/120 (8.3)	0.493 (0.204, 1.187)	0.114	0.538 (0.248, 1.166)	0.116	–0.073 (–0.164, 0.019)	0.118
Postoperative bleeding	13/94 (13.8)	6/120 (5.0)	0.323 (0.114, 0.916)	0.034	0.359 (0.139, 0.931)	0.035	–0.089 (–0.171, –0.006)	0.035
Total parental nutrition	17/83 (20.5)	21/112 (18.8)	1.015 (0.482, 2.138)	0.969	1.012 (0.558, 1.834)	0.969	0.002 (–0.114, 0.118)	0.969
Required intervention	15/94 (16.0)	18/120 (15.0)	0.917 (0.422, 1.991)	0.827	0.930 (0.488, 1.773)	0.827	–0.011 (–0.111, 0.089)	0.827

Values are presented as number (%).

CI, confidence interval.

Overall survival

During five years of follow-up, the mortality rate was slightly more significant in the myosteatosis group, with 0.82 deaths per 1,000 person-years, compared to 0.71 in the non-myosteatosis group. Adjusting for age and sex in a Cox regression, the overall risk of death was 19.2% greater in the myosteatosis, compared to the non-myosteatosis group. However, this was not statistically significant (p = 0.289) (Fig. 1). The median survival was 24 months within this cohort, with no significant difference between the myosteatosis and non-myosteatosis groups.

Fig 1. Kaplan–Meier Survival curves of both the myosteatosis and non-myosteatosis group patients.

This study demonstrates that in a cohort of patients who underwent operative treatment for suspected pancreatic and periampullary malignancy, the prevalence of myosteatosis was 56.1%, comparable to the published literature [10,17]. In a retrospective study of 218 patients with unresectable pancreatic cancers, Rollins et al. [10] found the prevalence of myosteatosis to be 55.3%. In a systematic review of 35 studies that reported CT analysis of myosteatosis, McGovern et al. [17] observed a 55.3% median prevalence in patients with pancreatic cancers. Except for gastric cancers (71.2%), the authors found that the prevalence of myosteatosis was similar in colorectal, esophageal, hepatobiliary, and breast and lung cancers at 52.1%, 54.3%, 56.8%, and 52.6%, respectively. Peng et al. [18] studied the impact of sarcopenia, myosteatosis, and sarcopenic obesity in patients with pancreatic cancer who underwent PD. The prevalence of myosteatosis was 39.7%, which is lower than our cohort, and may be due to the inclusion of periampullary cancers in our study.

To the best of our knowledge, this is the only study that has examined the impact of myosteatosis on both the short-term and long-term outcomes following pancreatic surgery for suspected malignancy. Some short-term complications, particularly the rate of the pancreatic leak and postoperative bleeding, were higher in the non-myosteatosis group. Surprisingly, although the overall postoperative bleeding rate was higher in the non-myosteatosis group, the blood transfusion rate was higher in the myosteatosis group. Although we have not reviewed the grading of the post-pancreatectomy hemorrhage, it is most likely that there were more patients with grade B & C PPH patients in the myosteaosis group, which would explain the higher rate of blood transfusion in the myosteatosis group, compared to the non-myosteatosis group. This result may reflect bias from unadjusted confounding, such as intrinsic differences in the care pathway for patients with myosteatosis.

While it may seem surprising that preoperative myosteatosis did not impact long-term survival within this cohort, we observed a non-significant higher death rate in the myosteatosis group. While this could be due to the small sample size and inclusion of pancreatic and periampullary cancers in the whole cohort, it may also be due to other contributing factors. Firstly, the 5-year survival for all patients undergoing surgery for pancreatic or periampullary malignancies remains low, with many studies quoting survival rates of less than 15%, in contrast to colorectal malignancy, which has an excellent 5-year survival following surgery, reported as over 80%. Therefore, the impact of nutritional status and associated frailty within a cohort with colorectal cancer is likely to become significant during extended survival, compared to the inherently poor survival in pancreatic malignancy [18]. In addition, we have included patients who have undergone rigorous preoperative assessment by focusing only on those undergoing surgery. PD is a large, complex, and time-consuming operation with a lengthy and high anesthetic risk, so preoperative assessment is already extensively performed using Cardiopulmonary Exercise Tolerance (CPET) testing. Therefore, those undergoing operative management will have been selected for better outcomes, and these patients are likely to be fitter and have better nutritional status. CPET has been demonstrated to correlate excellently with myosteatosis [19], removing from our cohort those likely to have poorer outcomes due to their nutritional status. It is well documented that oncological survival is significantly lower in those patients with myosteatosis in whole cohorts, irrespective of the decision to undergo operative management [9]. Previous literature has also established that myosteatosis can be detected on CT imaging before the onset of clinically detectable frailty [19]. When combining this with the overall low 5-year survival, it may also be the case that many patients are not surviving long enough for their myosteatosis to cause a functional decline.

Our results regarding the lack of impact of myosteatosis on long-term survival are similar to those of the only previously published paper in this area. Peng et al. [18] studied the effect of both myosteatosis and sarcopenia on pancreatic cancer survival following surgery. The authors found no association between myosteatosis and disease-free or overall survival. However, they found an association between sarcopenia and sarcopenic obesity and reduced survival in these patients. Therefore, future research is advised to assess both myosteatosis and sarcopenia, to determine the impact of these body composition modalities within the prognostication process.

The data obtained from our study of patients with operable pancreatic cancer appears to differ from those studies evaluating other gastrointestinal malignancies [7,17], which demonstrated increased postoperative complications in patients with myosteatosis in colorectal and oesophagogastric cancer. The meta-analysis from Lee and Kang [20] showed that myosteatosis is associated with significantly worse overall survival in patients with colorectal cancer, and advised that its measurement could be utilized in the prognostication process. Murnane et al. [7] also showed that myosteatosis is associated with a significantly increased risk of overall and severe complications, as well as substantially reduced long-term survival following radical oesophageal and gastric cancer surgery. A recent meta-analysis [21] demonstrated poorer long-term survival outcomes following surgery for all causes of gastrointestinal malignancy. As a result, additional research is required within the pancreatic cancer cohort to determine the potential use of myosteatosis, and why it may not be as associated with overall survival as with other gastric malignancies.

This study has several limitations. Firstly, it is a single-center retrospective study, with inherent weaknesses and bias. However, all patients in this cohort were operated on by the same group of surgeons, using similar reconstruction techniques. All postoperative complications were recorded prospectively. In addition, to avoid the impact of different types of pancreatic resections, we only included patients who underwent PD for malignant indications. A further area for improvement is that this data is from a single center, and the results may only apply to some of the population. Lastly, we combined four types of tumors with different long-term survival profiles in the study. Patients with pancreatic ductal adenocarcinoma have poor 5-year survival rates, compared to ampullary/duodenal/lower common bile duct cholangiocarcinoma. Note that no adjustment was made for multiple comparisons, so these unexpected results could potentially be an artefact of multiple testing, highlighting the constraints of exploratory analysis.

In conclusion, we observed a non-significant decrease in 5-year overall survival in the myosteatosis group, adjusting for age and sex. We also observed that myosteatosis was associated with a lower risk of POPF and PPH. We recommend a large prospective multi-center study to evaluate the impact of myosteatosis on the outcomes following PD for malignancy.

None.

No potential conflict of interest relevant to this article was reported.

Conceptualization: PJ, AM, MP, SA. Data curation: PJ, AM. Data analysis: AS. Methodology: PJ, AM, MP, SA. Writing - original draft: PJ, AM, GM, SA. Writing - review & editing: MP, GM, SM, AS, SA.