記住我
REVIEW ARTICLE
Year : 2022 | Volume
: 59
| Issue : 3 | Page : 310-316
A meta-analysis of prognostic factors in patients with left-sided pancreatic cancer
Adeleh Hashemi Fard1, Ramin Sadeghi2, Seyed Ehsan Saffari3, Seyed Majid Hashemi Fard4, Mohsen Aliakbarian1
1 Surgical Oncology Research Center, Mashhad University of Medical Sciences, Mashhad, Iran
2 Nuclear Medicine Research Center, Mashhad University of Medical Sciences, Mashhad, Iran
3 Centre for Quantitative Medicine, Duke-NUS Medical School, National University of Singapore, 8 College Road, Singapore
4 Department of Biology, Payame Noor University, Tehran, Iran
Correspondence Address:
Mohsen Aliakbarian
Surgical Oncology Research Center, Mashhad University of Medical Sciences, Mashhad
Iran
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/ijc.IJC_1150_20
Background: Patients with ductal adenocarcinoma of the body and tail of the pancreas usually remain asymptomatic until late in the course of the disease, and the survival of such patients depends on multiple factors, which may affect the therapeutic approach and patient survival. Hence, the aim of this study was to investigate such risk factors by pooling various available studies.
Methods: A systematic review was conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines between January 1, 2007, and December 31, 2016, using the following databases: Medline, Scopus, the Cochrane Library, and Google Scholar. Studies were selected according to the predesigned eligibility criteria, and information was extracted for demographics, clinical features, and survival outcomes. Data were pooled using fixed- or random-effects models.
Results: Sixteen studies were included (5,660 patients) with a median age of 64.8 years and a median survival of 28.5 (range 13–38) months. Identified significant factors for overall survival were higher age (hazard ratio [HR] = 1.211), men (HR = 1.182), presence of lymph node metastasis (HR = 1.964), multivisceral resection (HR = 1.947), N stage (1 versus 0; HR = 1.601), surgical margin (R0 versus No R0; HR = 0.519) and tumor size (>3 cm; HR = 1.890).
Conclusion: The pooled results of this study revealed several risk factors for overall survival in patients with left-sided pancreatic cancer.
Keywords: Cancer, distal, left sided, pancreatic, survival
Key Message Significant risk factors of worse overall survival are: Higher age, male gender, presence of lymph node metastasis, presence of multivisceral resection, R1 surgical margin, tumor size >3 cm.
Ductal adenocarcinoma of the body and tail of the pancreas usually remain asymptomatic until late in the course of the disease.[1],[2],[3],[4],[5] On diagnosis, the tumor size is generally larger in the pancreatic body and tail, and that is why it is less commonly resectable.[6]
Distal pancreatectomy (DP) is the treatment of choice for tumors of the left side of the pancreas with a clear margin and regional lymph node clearance for the patients without extensive peripancreatic lymphatic involvement, peritoneal dissemination, or distant metastases.[7],[8],[9],[10] In some institutes, patients with invasion of major vascular structure undergo distal pancreatectomy with vascular resection. However, the safety and efficacy as well as the survival outcome of such treatment are not clear.[11],[12],[13]
On the other hand, the only options for these advanced pancreatic cancers are chemo- and/or radiotherapies, but their effects are not very promising. The survival rate depends on whether undergoing surgery, the operation method, presence of positive nodes, and other clinical parameters. The 2-year survival rate is reported as 10% in unresectable pancreatic cancer with a median overall survival of 9.8 months[14], and the 5-year survival rate of DP with multimodal treatments is reported as 29% with a median overall survival of 35 months.[10] However, such data are limited.
Given that, in patients with tail and body pancreatic tumors, both those who have undergone surgery and those who have not been operated and instead have undergone chemo- and/or radiotherapies, information about the factors that affect their survival rate is not sufficient. Because of the time limit, it takes several years to conduct a large study to investigate such risk factors. Meta-analysis is an efficient approach to combine all the available studies for a better understanding of the risk factors, and a meta-analysis study, which is extensively based on various survival-related factors, has not been performed. Hence, the aim of this article was to study such risk factors by pooling various available studies to come up with more robust and consolidated results.
Materials and MethodsThis study was conducted based on the Preferred Reporting Items for Systematic and Meta-Analyses (PRISMA) guidelines (reference). The literature search, study selection, data extraction, and critical appraisal of the final selected studies were performed by two independent authors. As preplanned, any disagreement on eligibility for the study was resolved by discussion. A computerized search was made using Medline, Scopus, the Cochrane Library, and Google Scholar. Publications between January 2007 and December 2016 were included, and December 31, 2019, was the ultimate search date limited to the English language. The keywords for the search were “pancreatic AND cancer AND survival AND (left-sided OR distal).” There was no need for ethical approval because this literature review had no deleterious effects on the patients.
Inclusion criteria were studies with distal pancreatic cancer patients, observational retrospective/prospective study design, and time to death outcome. Exclusion criteria were studies published in non-English languages; published as abstracts, letters, comments, editorials, expert options, and reviews without original data; on nonhuman participants; duplicate studies with the same center; and insufficient data on primary outcome. The most relevant (or recent) study was included in case there were more than one overlapped studies.
After the duplicate studies were removed, two independent authors checked the title and abstract (and full-text for detailed evaluation) of the remaining studies for adherence according to the inclusion and exclusion criteria. The two authors extracted the information from the shortlisted studies. The detailed information were first author, article title, year of the publication, sample size of the study, baseline characteristics, clinical features, and time to death; predefined data extraction forms were used to collect the above variables. The methodological quality and risk of bias of the final shortlisted studies were evaluated using the Cochrane Handbook for Systematic Reviews of Interventions (reference). We assessed the publication bias via funnel plots and tested by using Egger's test and Begg's test.
All statistical analysis was performed using Stata/SE Version 13.1 for Windows (StataCorp LLC, College Station, TX, USA). Hazard ratio (HR) and 95% confidence intervals (CIs) were calculated using fixed- or random-effects models to investigate the potential risk factors of the survival outcome. Statistical heterogeneity among the studies was computed by I2 statistics for each parameter. The presence of statistical heterogeneity was considered significant if I2 was >50%. To combine the results, fixed- and random-effect models were used for the presence and absence of heterogeneity scenarios, respectively. Statistical significance was set at P < 0.05.
ResultsSixteen studies[9],[10],[15],[16],[17],[18],[19],[20],[21],[22],[23],[24],[25],[26],[27],[28] were obtained via a systematic search fulfilling the eligibility criteria [Figure 1], which included 5,660 patients between 1990 and 2015. The case-weighted median age is 64.8 years. Country-wise, four studies were conducted in Japan,[15],[16],[19],[27] four in South Korea,[20],[21],[23],[24] three in the United States,[17],[18],[22] one in Italy,[25] two in France,[10],[26] one in China,[28] and one in the Netherlands.[9] Most of the studies were retrospective (13 out of 16). The mean duration of the included studies is 12.1 ± 5.4 years (range: 5–22 years). Median overall survival was 28.5 months (range: 13–38 months) [Table 1].
Age was reported in five studies,[9],[10],[15],[17],[25] which included 2,166 patients. Heterogeneity of this factor was not significant among the studies (I2 = 22.8%, P = 0.269). Results of fixed-effect meta-analysis show that higher age is a significant predictor for a worse overall survival (HR: 1.211, 95% CI [1.063, 1.38], P = 0.004) [Supplement Figure 1].
Gender of the individuals was available in six studies,[9],[10],[15],[17],[25],[26] including 4,919 patients. Heterogeneity was not statistically significant (I2 = 0%, P = 0.722). Significant association was found between the patients' gender and overall survival (better survival outcome for women) using fixed-effect meta-analysis (HR [men versus women): 1.182, 95% CI [1.083, 1.291], P < 0.001) [Supplement Figure 2].
Six studies[15],[16],[18],[19],[23],[24] including 448 patients reported the presence or absence of lymph node metastasis. Heterogeneity was not significant (I2 = 49.8%, P = 0.063). Meta-analysis showed that the presence of lymph node metastasis was significantly associated with worse overall survival (HR: 1.964, 95% CI [1.678, 2.298], P < 0.001) [Supplement Figure 3].
Adjuvant therapy was reported in seven studies[9],[14],[15],[22],[24],[25],[28] with a total patient number of 695. Significant heterogeneity was observed (I2 = 87.9%, P < 0.001). Presence of adjuvant therapy was not significantly associated with the overall survival using random-effect meta-analysis (HR: 1.284, 95% CI [0.672, 2.456], P = 0.449) [Supplement Figure 4].
Three studies[9],[10],[22] with a total number of patients of 654 reported the American Society of Anesthesiologists (ASA) grade. No significant heterogeneity was observed (I2 = 0%, P = 0.493). ASA grade was not significantly associated with overall survival (HR: 1.278, 95% CI [0.956, 1.707], P = 0.097) [Supplement Figure 5].
Five studies[10],[15],[16],[24],[28] including 576 patients reported histologic grade. Significant heterogeneity was observed for this parameter (I2 = 77.9%, P < 0.001). Random-effect meta-analysis indicates that the association between histologic grade and overall survival was not significant (HR: 0.88, 95% CI [0.396, 1.954], P = 0.753) [Supplement Figure 6].
Data on multivisceral resection were available from 442 patients from two studies[9],[10]. Heterogeneity was not statistically significant (I2 = 0%, P = 0.678). Presence of multivisceral resection was significantly associated with worse overall survival (HR: 1.947, 95% CI [1.371, 2.764], P < 0.001) [Supplement Figure 7].
Node (N) stage data were available from seven studies[9],[10],[15],[16],[22],[25],[28] with 924 patients. Heterogeneity was not significant (I2 = 39.2%, P = 0.131). Higher N stage was significantly associated with worse overall survival (HR [N stage 1 versus 0]: 1.601 95% CI [1.273, 2.013], P < 0.001) [Supplement Figure 8].
Five[9],[22],[23],[26],[28] studies (3,247 patients) reported the operation procedure (laparoscopic distal pancreatectomy [LDP] versus open distal pancreatectomy [ODP]). Significant heterogeneity was observed for this parameter (I2 = 65.6%, P = 0.020). LDP and ODP did not show a significant different in overall survival using random-effect meta-analysis (HR: 0.994, 95% CI [0.612, 1.614], P = 0.980) [Supplement Figure 9].
Post-op complications data were described in two studies[9],[23] with 231 patients. Heterogeneity of this parameter was significant (I2 = 88.6%, P = 0.003). Results of random-effect meta-analysis show no significant association between the post-op complications and overall survival (HR: 0.89, 95% CI [0.217, 3.657], P = 0.872) [Supplement Figure 10].
R0 (versus No R0) surgical margin data were reported in 714 patients from eight studies.[16],[19],[20],[22],[23],[25],[27],[28] Heterogeneity of surgical margin was found to be not statistically significant (I2 = 7%, P = 0.376). R0 surgical margin was positively associated with a better overall survival (HR: 0.519, 95% CI [0.391, 0.69], P < 0.001) using fixed-effect meta-analysis [Supplement Figure 11].
Data on tumor (T) stage were available for 270 patients from four studies.[15],[16],[25],[28] Heterogeneity was not significant (I2 = 0%, P = 0.465). There was no significant difference in overall survival between T stage of >2 versus 1–2 (HR: 1.333, 95% CI [0.751, 2.364], P = 0.326) [Supplement Figure 12].
Studies reported tumor size in different ways. Two studies with 160 patients[25],[27] and another two studies with 513 patients[10],[22] reported a cutoff point of 3 cm and 4 cm for tumor size, respectively. Heterogeneity for tumor size >3 cm was not significant (I2 = 5%, P = 0.305); however, this statistic for tumor size >4 cm was more than 50% (I2 = 69.4%, P = 0.071). Significant association was found between tumor size >3 cm and worse overall survival (HR: 1.89, 95% CI [1.175, 3.041], P = 0.009). However, the cutoff point of 4 cm was not found to be significantly associated with overall survival (fixed-effect meta-analysis: HR: 1.357, 95% CI [0.961, 1.917], P = 0.083; random-effect meta-analysis: HR: 1.222, 95% CI [0.616, 2.425], P = 0.566) [Supplement Figure 13] and [Supplement Figure 14].
Data on the World Health Organization (WHO) grading were available from two studies[17],[25] with 1,560 patients. Heterogeneity was observed to be significant (I2 = 77.9%, P = 0.033). WHO grading (Grades 3–4 versus Grades 1–2) was not significantly associated with worse overall survival (HR: 2.188, 95% CI [0.516, 9.279], P = 0.288) [Table 2] [Supplement Figure 15].
Two studies reported lymph node ratios: one as a numeric parameter (HR: 2.94, 95% CI [0.95, 9.12], P = 0.062)[21] and the other as binary (HR [>0.1 versus ≤1]: 1.383, 95% CI [0.097, 1.742], P = 0.006).[17] Perineural invasion is reported in two included papers with HR (negative versus positive) of 2.89 (95% CI [1.24, 6.75], P = 0.014)[28] and HR (positive versus negative) of 1.98 (95% CI [0.59, 12.33], P = 305).[15] Vascular invasion was presented in two included studies as positive versus negative with HR: 3.183 (95% CI [1.406, 7.228], P = 0.006),[15] and negative versus positive with HR: 2.648 (95% CI [1.157, 6.301], P = 0.0209).[16]
DiscussionSeveral risk factors (including demographics and baseline clinical features) are found to be associated with overall survival in left-sided or distal pancreatic cancer patients. Most studies included in this meta-analysis show that higher age is significantly associated with worse overall survival (HR: 1.211, 95%CI [1.063, 1.38], P = 0.004). This is an expected result as underlying disease such as respiratory and cardiovascular diseases is more common in the elderlies. Also, because of the weaker immunity system in older patients, less tolerance to surgery and post-op chemotherapy is expected. The results of the current study show that a better overall survival outcome was found in women compared with men (HR (men versus women): 1.182, 95% CI [1.083, 1.291], P < 0.001). This could be because of lower age of illness, seeing a doctor faster due to onset of pain, lower proportion of drugs/opium users in women, which would lead to diagnosis in a lower severity of the disease. Studies included in the current article indicate that lymph node metastasis is significantly associated with worse overall survival, which is anticipated (HR: 1.964, 95% CI [1.678, 2.298], P < 0.001). The reason might be due to metastasis of pancreatic cancer via lymph that would translate to a higher grade.
The current study also shows that the presence of adjuvant therapy was not significantly associated with the overall survival (HR: 1.284, 95% CI [0.672, 2.456], P = 0.449). Seven studies included in the meta-analysis of adjuvant therapy used different therapies: four chemotherapy, two chemotherapy or chemoradiotherapy, and one chemoradiotherapy. The high heterogeneity for adjuvant therapy (I2 = 87.9%, P < 0.001) may be due to methodological diversity in various studies. Further investigations are needed on the type of medicine and dosage in the adjuvant therapy to conclusively examine the overall survival for such patients. Results of the three studies included with reported ASA grade showed that ASA grade was not significantly associated with overall survival (HR: 1.278, 95% CI [0.956, 1.707], P = 0.097). Our results indicated that the presence of multivisceral resection was found to be significantly associated with worse overall survival (HR: 1.947, 95% CI [1.371, 2.764], P < 0.001). Higher grade and higher risk of complications such as thrombosis bleeding might be potential reasons. Also, multivisceral resection may result in ischemia of the hepatoduodenal region, which would cause ischemic gastropathy and ischemic hepatopathy.
In the current study, a higher N stage was found to be a significant predictor of worse overall survival (HR [N stage 1 versus 0]: 1.601, 95% CI [1.273, 2.013], P < 0.001). Further expansion to the lymph node is more likely to be the case in higher N stage, which would lead to increase the chance of metastasis and worsen the survival. Six studies included in the meta-analysis of operation procedure show that LDP and ODP did not significantly differ in overall survival. Hence, operation procedure may not be a critical decision from the survival point of view. Three studies reported that the post-op complications and the pooled results show that the post-op complications are not significantly associated with overall survival. A high heterogeneity (I2 = 65.6%, P = 0.02) is calculated for the post-op complications because the definition of the post-op complications may not be consistent in the different studies that were included. The current study shows that the R0 surgical margin was significantly associated with a better overall survival (HR: 0.519, 95% CI [0.391, 0.69], P < 0.001). This is an anticipated finding as the entire tumor is resected, leading to a less chance of metastasis. The results of the studies that reported T stage to indicate that the T stage is not significantly associated with overall survival. A potential reason for this might be the metastasis via lymph node. Four included studies reported tumor size. Out of these four, cutoff points of tumor size >3 cm and tumor size >4 cm were reported in two studies each. Although the pooled HRs show that higher tumor size leads to worse overall survival (pooled HR >1), as expected, only the cutoff point of tumor size >3 cm (versus <3 cm) was found to be significantly associated with worse overall survival (HR: 1.89, 95% CI [1.175, 3.041], P = 0.009). From a mathematical point of view, a greater cutoff point may lead to a lower sensitivity, hence the association signal might be weak for cutoff point of tumor size >4 cm. No significant association was found between histologic grade and overall survival in the current study. All five studies included in the meta-analysis of histologic grade reported the tumor differentiation as well/moderate versus poor. Some literature reported the histologic grading as the WHO grading. Pooled results of four included studies indicate that WHO grading (defined as Grades 3–4 versus Grades 1–2) was not significantly associated with overall survival (P = 0.288). The heterogeneity was pretty high for histologic grade parameter (I2 = 77.9%, P< 0.001) and WHO grading (I2 = 77.9%, P = 0.033) because various studies may have used different methods that are not reported in these studies; it does not seem that histologic grade is an independent predictor of the overall survival. Hence, this may not be a significant parameter to consider for surgery.
Finally, the significance of the presence of lymph node metastasis, N stage, and R0 surgical margin along with their vascular involvement should be noted, which are all associated with metastasis and leading to increasing the risk of metastasis. The clinical question of this study has not been investigated previously. Some of the clinically important parameters, found to be associated with worse survival in this study, were also discussed in previous studies. The findings of the current study tallies with similar studies,[29],[30] which elaborated the importance of R0 and vascular involvement. Previous systematic review and meta-analysis papers discussed the survival of patients who underwent distal pancreatectomy with en bloc celiac resection (DP-CAR). Because such patients would have vascular involvement and undergo DP-CAR due to R0 and resection, better survival in DP-CAR would lead to better survival with R0 resection. It is also reported that some patient characteristics are associated with better survival, including age <60 years and microscopically radical tumor resection. It has been discussed that DP-CAR could increase the R0 rate to 72.79% with a high incidence of vascular reconstruction.[29],[30]
The current study has several limitations. Most of the included studies were retrospective with a potential high risk of selection bias. The small sample size of these studies (11 studies with a sample size of <100 patients) is the other limitation, which would lead to a low level of clinical evidence. This is a research challenge for surgery cases that are not very common, because it may take several years if one would like to study a larger sample size. The average duration of the included studies was 12.1 years (minimum–maximum: 5–22 years). Hence, various treatment and management protocols over time may affect surgical outcomes and survival consequently.
In conclusion, the pooled results of this study revealed several risk factors for overall survival in patients with left-side pancreatic cancer. However, because of the above-mentioned limitations and controversial issues, the conclusions should be interpreted with caution.
Financial support and sponsorship
Nil.
Conflicts of interest
References
留言 (0)