Recently, the value of computed tomography (CT)-derived body composition measures for predicting postoperative outcomes has gained interest. For example, loss of skeletal muscle mass shown on preoperative CT imaging has been associated with worse short- and long-term outcomes after resection of tumors of varying origin.
1 Also in gastrointestinal and hepato-pancreato-biliary (HPB) surgery,
2 including surgery for pancreatic cancer,
3‐
6 low skeletal muscle mass has been associated with increased morbidity and mortality and worse survival. Less consideration has been given to muscle attenuation shown on CT imaging, which has been associated with worse survival for patients with solid tumors of respiratory and gastrointestinal origin, for patients with melanoma or metastatic renal cell carcinoma,
7‐
9 and recently for patients with pancreatic cancer.
4,
5,
10
Studies associating preoperative CT-derived body composition measures of patients having periampullary, nonpancreatic cancer with survival and major postoperative complications do not exist. However, periampullary cancer constitutes about one third of all patients undergoing pancreatoduodenectomy (PD).
15 The 5-year survival rate after resection of periampullary (distal bile duct, papilla, and duodenum) cancer may reach 50 %, twice the rate for pancreatic cancer patients.
15,
16 Mortality rates after PD are about 1 to 3 % in expert centers, but postoperative morbidity occurs in up to 50 % of patients.
15,
17 Therefore, risk stratification and response prediction before treatment remain of great interest. We assessed the association of skeletal muscle mass and muscle attenuation (quality) with OS and major postoperative complications in patients undergoing PD for periampullary, nonpancreatic cancer.
Discussion
In this first study investigating CT-derived body composition measures in patients undergoing PD for periampullary, nonpancreatic cancer, low muscle quality, but not low muscle mass, was independently associated with decreased survival and an increased incidence of major postoperative complications.
Various well-established patient and tumor characteristics are known to affect short- and long-term postoperative outcomes for patients with pancreatic and periampullary cancer.
16,
23‐
26 Our study adds to the previously identified risk factors for these patients, with a prognostic value equalling that of other conventional covariates.
The recent international interest in CT-derived body composition measures has led to identification of muscle mass and muscle quality as new independent predictors of outcome, with growing evidence to validate their prognostic impact.
2 After some very recent studies investigating pancreatic cancer,
3‐
6,
10 our study is a first step toward determining the use of preoperative CT-derived body composition measures to stratify all patients receiving PD for cancer into risk categories and to assist in shared decision making.
Ideally, treatment of low muscle quality would lead to improved postoperative outcomes. Unfortunately, no such treatment is currently available.
1 A randomized controlled trial with 52 nonsurgical, elderly patients demonstrated the clear effects of moderate physical exercise on age-related decline in muscle strength, but it involved a 12-month training program of moderate physical activity
29 and therefore has no relevance in the preoperative setting or during neoadjuvant treatment. However, several prehabilitation strategies consisting of physical exercise have demonstrated improved postoperative outcomes.
30 The effect of preoperative muscle or total-body exercises on postoperative outcomes for patients with low muscle mass or low muscle quality should be subject to future investigation.
Until recently, very little research on skeletal muscle attenuation (quality) was conducted. Some recent studies have investigated the prognostic significance of low muscle attenuation for patients with pancreatic cancer.
4,
5,
10 One study included 230 patients undergoing PD, distal pancreatectomy, or total pancreatectomy for pancreatic ductal adenocarcinoma and identified muscle attenuation as an independent predictor of both worse survival and recurrence-free survival.
4 In another study of 104 pancreatoduodenectomies, skeletal muscle attenuation was an independent predictor of severe complications according to the National Surgical Quality Improvement Program (NSQIP).
5 It was not described whether periampullary cancer in the final pathology diagnosis was included in this study. Recently, attenuation of the psoas muscle on preoperative CT scan also was identified as an independent predictor of 1-year mortality for 518 elderly patients receiving HPB surgery, including pancreatic resection.
10 In the limited number of studies available, skeletal muscle quality was an important covariate in the prognosis of patients receiving PD for cancer.
Previous studies have identified muscle mass as an independent predictor of postoperative complications and survival and also after resection of pancreatic cancer.
1‐
6 In one of the earlier, large series after resection of pancreatic adenocarcinoma, muscle mass was not associated with the risk of overall morbidity or serious complications but was associated with an increased risk of 3-year mortality.
3 To our surprise, we did not find an association between muscle mass and postoperative major complications or survival in the univariate analysis. It is unlikely that the influence of muscle mass on postoperative outcomes is different after resection of periampullary cancer compared with pancreatic cancer, although the incidence of complications does differ between the two groups.
28,
31,
32 It is unclear whether the unequal distribution of patients between low muscle mass (78 %) and normal muscle mass (22 %) could be a reason for this finding.
We used the preferred and accepted method (optimal stratification) to find the most significant cutoff for L3 muscle index, as previously described.
1,
21 We found different cutoffs for men and women, which suggests that other, “static” cutoffs do not suit our cohort, so we believe that using these cutoffs would lead to inaccurate effect measurement. However, also when other previously used cutoff values for low muscle mass were applied, muscle mass still showed no effect on survival.
2 The patients were better distributed among the muscle quality categories (51 vs 49 %). This suggests that some patients had a low muscle mass, but with normal muscle quality.
We found a significant negative correlation between MAI and intramuscular adipose tissue on preoperative CT scan, which reflects an accumulation of intramuscular lipid content.
11‐
13 A higher muscle lipid content has been related to loss of muscle strength.
14 This may explain why muscle quality and not muscle mass was an independent predictor of survival and major postoperative complications.
Complications after pancreatic surgery are very common.
17,
19,
20 After resection for periampullary cancer, complications may occur even more frequently than after resection for pancreatic adenocarcinoma.
28,
31,
32 The incidence of overall and major morbidity did not differ between the patients with low and normal muscle mass. The patients with low muscle quality experienced overall and major complications more frequently than the patients with normal muscle quality. These findings are reflected in our identification of low muscle quality as an independent predictor of survival and major complications.
We used several of the strongest and most frequent covariates predicting postoperative outcomes after pancreatic cancer surgery.
16,
23‐
26 Unfortunately, the presence of perineural or perivascular invasion was documented only incidentally by the pathologist and could therefore not be incorporated into our analysis.
We found a relatively long interval (median 55 days) between the final preoperative CT scan and the date of surgery. Possibly, as patients are referred to our tertiary hospital including CT scan, the interval is increased due to first discussion of the patient within a multidisciplinary team meeting, after which the operation is planned. Findings have shown that scans performed at a longer interval before surgery increases the incidence of intraoperative metastatic disease.
33 Patient and event numbers were unfortunately too small to determine whether results would be different if patients who had a scan performed more than 30 days before the surgery were excluded.
This study had some limitations. We had to exclude a relatively high number of patients due to inadequacy of CT scans. However, the baseline characteristics did not differ significantly between the included and excluded patients (Supplementary Table 1). Furthermore, we identified some differences in baseline characteristics, mainly, a lower R1 rate for the patients with a low muscle mass and a higher BMI for the patients with low muscle attenuation than for the other patients. However, we corrected for these and other possible confounders in our analyses. Finally, we included three periampullary cancer types in our cohort. Survival and the incidence of complications may differ between the three types of periampullary tumors (duodenum, papilla, and distal bile duct tumors).
16,
28 However, in the sensitivity analysis, including also the cancer type in the multivariable analyses did not influence survival or the rate of major complications (data not shown).