Abstract
Background/Objectives:
The aim of this study was to determine the role of site-specific metastatic patterns over time and assess factors associated with extended survival in metastatic PDAC. Half of all patients with pancreatic ductal adenocarcinoma (PDAC) present with metastatic disease. The site of metastasis plays a crucial role in clinical decision making due to its prognostic value.
Methods:
We examined 56,757 stage-IV PDAC patients from the National Cancer Database (2016–2019), categorizing them by metastatic site: multiple, liver, lung, brain, bone, carcinomatosis, or other. The site-specific prognostic value was assessed using log-rank tests while time-varying effects were assessed by Aalen’s linear hazards model. Factors associated with extended survival (>3years) were assessed with logistic regression.
Results:
Median overall survival (mOS) in patients with distant lymph node-only metastases (9.0 months) and lung-only metastases (8.1 months) was significantly longer than in patients with liver-only metastases (4.6 months, p<.001). However, after six months, the metastatic site lost prognostic value. Logistic regression identified extended survivors (3.6%) as more likely to be younger, Hispanic, privately insured, Charlson-index <2, having received chemotherapy, or having undergone primary or distant site surgery (all p<.001).
Conclusion:
While synchronous liver metastases are associated with worse outcomes than lung-only and lymph node-only metastases, this predictive value is diminished after six months. Therefore, treatment decisions beyond this time should not primarily depend on the metastatic site. Extended survival is possible in a small subset of patients with favorable tumor biology and good conditional status, who are more likely to undergo aggressive therapies.
Keywords: Pancreatic ductal adenocarcinoma, metastatic disease, sites of metastasis, extended survival, pancreas
INTRODUCTION
Pancreatic ductal adenocarcinoma (PDAC) continues to have one of the highest mortalities of any type of cancer. The 5-year survival for patients with resectable and borderline resectable PDAC is reported to be 44%, followed by locally advanced PDAC at 15%, and metastatic disease at only 3%.1 Only a modest improvement in survival has been observed over the last decade.2 One of the major reasons for this dismal prognosis is the high propensity for early dissemination of disease.3, 4 Additionally, the often asymptomatic nature of early-stage disease results in approximately 50% of patients having metastatic disease at diagnosis.5,6
While systemic therapy is the standard of care for patients metastatic PDAC, curative-intent management has demonstrated safety and efficacy in highly selected patients.7–9 Retrospective studies suggest favorable survival compared to historical data after a response to induction therapy and resection of oligometastatic disease when performed at specialized centers.10, 11
Due to the lack of effective biomarkers to appropriately stratify patients with favorable disease biology, multiple factors have been investigated to help identify patients that may benefit from a curative-intent approach.12, 13 The site of metastasis is an important factor to consider when evaluating patients with metastatic PDAC.14–16 The liver is the most common site of initial distant metastasis with frequently progression to one or multiple other sites such as lungs, brain, bone, adrenal glands, and peritoneum.17 Recent retrospective studies have demonstrated differential survival depending on the metastatic site.18–21 Specifically, patients with lung-only metastases have significantly longer survival compared to those with liver metastasis.15, 21, 22 Another recent study looked at patterns of metastasis in different organs and similarly found that patients with liver-only metastases had significantly worse survival compared to those with lung-only metastases or distant lymph node-only metastases.19
With metastatic PDAC having such dismal outcomes there has been limited study of patients who survive for an extended period of time after diagnosis. Findings of small cohort studies have shown that lower CA19–9 and bilirubin levels at diagnosis as well as the absence of peritoneal metastasis are associated with longer survival.23–25
While the site of metastasis plays a crucial role in clinical decision making due to its prognostic value, no study has assessed the long-term impact of the site of metastases on outcomes. Therefore, we sought to determine the impact of site-specific metastasis on survival in short-term and extended survivors and identify clinical factors associated with extended survival in patients with metastatic PDAC.
METHODS
Study population
The National Cancer Database (NCDB) is a comprehensive clinical oncology database that collects data from over 1500 Commission on Cancer (CoC)-accredited institutions across the United States. The NCDB Pancreas participant user file (PUF) was used to identify all patients who presented with stage IV PDAC at diagnosis between 2016 and 2019. The ICD-O-3 codes 25.9 and 25.3 were included in the full data set. Institutional review board approval was not required as the NCDB is an administrative, deidentified data set.
Patients who had localized PDAC according to the NCDB Analytic Stage Group were excluded. The analytic stage was determined by pathologic stage group first, and, if unavailable, clinical stage group was used. Patients who had previous cancer diagnoses were also excluded from our analysis. Patients without distant metastasis who may have been miscoded in the database were also excluded. The patient selection is presented in Figure 1 and other key components of the cohort derivation are described in Supplemental Material Table 1.
Figure 1.
Exclusion cascade
Patient characteristics, including age at diagnosis, sex, race, ethnicity, income, insurance status, and Charlson-Deyo Score, were extracted from the NCDB file. Age was dichotomized: <65 years old and ≥ 65 years old. Race was collapsed into the category of white and non-white due to adequate numbers needed in each category. Income quartiles were based on equally proportioned income ranges among all US zip codes in the NCDB. Insurance status was determined by primary payer at diagnosis. Performance status was assessed using the Charlson-Deyo Index.
The primary exposure variable was site of metastatic disease. Patients were classified by site of metastatic disease at the time of diagnosis: multiple, liver, lung, brain, bone, or carcinomatosis. Furthermore, patients were designated as multiple sites of metastasis if they had >1 site of metastasis or as “other” if their distant metastasis was localized at a site different to the aforementioned.
Survival time of all patients was extracted along with whether the patient underwent surgical resection of the primary tumor site and whether the patient underwent surgery at a distant site. Patients were divided into two groups based on survival time. Short-term survivors were defined as survival of less than three years after diagnosis and long-term survivors were defined as living three or more years after diagnosis. We decided to set a cut off for “extended survivorship” of three years (3.6%) based on the closest applicable value for statistical outliners (μ±2SD) and clinical meaningfulness.
Statistical Analysis
Descriptive statistics were reported using mean and standard deviation, or median and range for continuous variables and frequency and percentage for categorical variables. Extended (long-term) survivors (≥3 years) were compared to short-term survivors for demographic, clinical, and pathological characteristics using Chi-squared test for categorical variables and Mann Whitney U tests for continuous variables.
Logistic regression was performed to predict extended survivors using statistically significant variables based off the univariate analysis in order to model whether patients survived greater than 3 years with adjustment for co-variates. Overall survival (OS) was calculated from time of diagnosis to time of death. If patients were lost to follow up, they were censored at time of last contact. Kaplan-Meier analysis was conducted with four patient groups defined by metastatic sites at different time points. The groups were: liver-only metastases, lung-only metastases, distant node-only metastases, and multiple organ-only metastases. The time points for survival were set at >3 months, >6 months, >1 year, >2 years and >3 years from diagnosis. Univariate Cox regression was performed for OS with the following covariates: age (greater than 65 years old or less than or equal to 65 years old), sex, race (white or non-white), ethnicity (Hispanic or Non-Hispanic), insurance status (not insured, private insurance, Medicaid, Medicare, or other government), sites of metastasis (liver, lung, distant lymph node, bone, brain, carcinomatosis, other or multiple organ metastasis), chemotherapy (no chemotherapy or any chemotherapy), surgery of primary site (no surgery or any surgery), and distant site surgery (no distant site surgery or any distant site surgery). Multivariate Cox regression was performed using all statistically significant variables from the univariate analysis.
Aalen’s linear hazards model was employed to investigate time-varying effects in proportional hazard regression concerning liver, peritoneal, and lung as metastatic sites. As reported by Hosmer et al.26 the expected behavior of the plot is interpreted in the following manner: if the time-varying coefficient for a covariate in Aalen’s model is constant, an increasing slope, of the plot of cumulative regression coefficient versus time, represents an increase in the hazard of death due to the covariate, while decreasing slope suggests a decrease in hazard of death. If the covariate has no further effect on the hazard of death, the plot is constant and parallel to the x-axis. Time dependence was assessed using the Cramér-von Mises to examine the time-invariance of the prognostic effect. A p-value of less than 0.05 was used to determine statistical significance.
RESULTS
Study population
There was a total of 161,751 patients in the NCDB Pancreas PUF diagnosed with pancreatic tumors between 2016 and 2019. Within these patients, 73,218 (45.3%) presented with stage IV PDAC. Of the 54.7% of patients who presented with localized disease approximately 17% had stage I disease, 21% had stage II disease and 11% had stage III disease.
There were 12,429 (7.7%) patients who had a previous cancer diagnosis and were therefore excluded and 4,032 (2.5%) patients who did not have information on metastatic site. A total of 56,757 (35.1%) patients with metastatic stage IV PDAC were identified and included for final analysis.
The median age of the entire cohort was 68 years (IQR 60–76). The cohort was majority white (81.3%) and male (54.0%). The majority of patients presented with liver metastases (51.5%) followed by multiple sites (31.2%), “other” (9.4%), lung (4.9%), distant lymph node (1.6%), bone (0.9%), carcinomatosis (0.6%), and brain (0.1%). A total of 54.7% of patients received chemotherapy, 2.6% had surgery of the primary tumor site, and 2.6% had a surgery of a distant site. (Table 1) Within the patients who had surgery of their primary tumor site, 654 (44.4%) had surgery less than a month after diagnosis, 479 (32.5%) had surgery between 1 month and 4 months after diagnosis and 340 (23.1%) had surgery 4 or more months after surgery.
Table 1:
Univariate Analysis of Short Survivors and Extended Survivors
| Total N=56757 n (%) | Short Survivor (Less than 3 years) N = 54705 n (%) | Extended survivor (Greater than 3 years) N = 2052 n (%) | P values | |
|---|---|---|---|---|
|
| ||||
| Age Category | <.001 | |||
| <65 Years Old | 21767 (38.4) | 20609 (37.7) | 1158 (56.4) | |
| ≥65 Years Old | 34990 (61.6) | 34096 (62.3) | 894 (43.6) | |
| Race | 0.047 | |||
| White | 45758 (81.3) | 44154 (81.3) | 1604 (79.6) | |
| Non-White | 10551 (18.7) | 10139 (18.7) | 412 (20.4) | |
| Sex | 0.027 | |||
| Male | 30677 (54) | 29617 (54.1) | 1060 (51.7) | |
| Female | 26080 (46) | 25088 (45.9) | 992 (48.3) | |
| Ethnicity | <.001 | |||
| Non-Spanish/Non-Hispanic | 51896 (93) | 50080 (93.1) | 1816 (90.7) | |
| Hispanic | 3884 (7) | 3698 (6.9) | 186 (9.3) | |
| Median Income Quartiles | <.001 | |||
| < $46,277 | 8352 (17.5) | 8097 (17.6) | 255 (14.6) | |
| $46,277 – $57,856 | 10321 (21.6) | 10016 (21.7) | 305 (17.4) | |
| $57,857 – $74,062 | 11566 (24.2) | 11169 (24.2) | 397 (22.7) | |
| ≥ $74,063 | 17597 (36.8) | 16804 (36.5) | 793 (45.3) | |
| Primary Payor at Diagnosis | <.001 | |||
| Not Insured | 1728 (3.1) | 1662 (3) | 66 (3.2) | |
| Private Insurance | 17032 (30.4) | 16088 (29.4) | 944 (46) | |
| Medicaid | 4257 (7.6) | 4084 (7.5) | 173 (8.4) | |
| Medicare | 32199 (57.5) | 31366 (57.3) | 833 (40.6) | |
| Other Government | 767 (1.4) | 751 (1.4) | 16 (0.8) | |
| Any Chemotherapy | <.001 | |||
| No Chemotherapy | 25180 (45.3) | 24393 (45.6) | 787 (39.3) | |
| Yes, received chemotherapy | 30369 (54.7) | 29154 (54.4) | 1215 (60.7) | |
| Any Surgery Distant Site | <.001 | |||
| No distant surgery | 54813 (97.4) | 53034 (97.7) | 1779 (89.5) | |
| Distant site surgery (Distant LN or other location) | 1448 (2.6) | 1240 (2.3) | 208 (10.5) | |
| Any Surgery | <.001 | |||
| No Surgery | 55107 (97.4) | 53456 (98) | 1651 (80.7) | |
| Any Surgery of Primary Site | 1484 (2.6) | 1090 (2) | 394 (19.3) | |
| Mets at Diagnosis - Multiple Organs | <.001 | |||
| No | 39055 (68.8) | 37420 (68.4) | 1635 (79.7) | |
| Yes | 17702 (31.2) | 17285 (31.6) | 417 (20.3) | |
| Mets at Diagnosis - Only Other (other than bone, brain, liver, lung or distant lymph nodes) | 0.874 | |||
| No | 51448 (90.6) | 49590 (90.6) | 1858 (90.5) | |
| Yes | 5309 (9.4) | 5115 (9.4) | 194 (9.5) | |
| Mets at Diagnosis - Only Carcinomatosis | 0.016 | |||
| No | 56419 (99.4) | 54371 (99.4) | 2048 (99.8) | |
| Yes | 338 (0.6) | 334 (0.6) | 4 (0.2) | |
| Mets at Diagnosis - Only Distant Lymph Node | <.001 | |||
| No | 55873 (98.4) | 53873 (98.5) | 2000 (97.5) | |
| Yes | 884 (1.6) | 832 (1.5) | 52 (2.5) | |
| Mets at Diagnosis - Only Bone | 0.360 | |||
| No | 56254 (99.1) | 54224 (99.1) | 2030 (98.9) | |
| Yes | 503 (0.9) | 481 (0.9) | 22 (1.1) | |
| Mets at Diagnosis - Only Brain | 0.125 | |||
| No | 56704 (99.9) | 54656 (99.9) | 2048 (99.8) | |
| Yes | 53 (0.1) | 49 (0.1) | 4 (0.2) | |
| Mets at Diagnosis - Only Liver | <.001 | |||
| No | 27549 (48.5) | 26726 (48.9) | 823 (40.1) | |
| Yes | 29208 (51.5) | 27979 (51.1) | 1229 (59.9) | |
| Mets at Diagnosis - Only Lung | 0.002 | |||
| No | 53997 (95.1) | 52075 (95.2) | 1922 (93.7) | |
| Yes | 2760 (4.9) | 2630 (4.8) | 130 (6.3) | |
| Charlson-Deyo Score | <.001 | |||
| 0 | 36328 (64) | 34844 (63.7) | 1484 (72.3) | |
| 1 | 12253 (21.6) | 11855 (21.7) | 398 (19.4) | |
| 2 | 4143 (7.3) | 4054 (7.4) | 89 (4.3) | |
| ≥ 3 | 4033 (7.1) | 3952 (7.2) | 81 (3.9) | |
Survival Based on Site of Metastasis
Unstratified log-rank test shows that patients with distant lymph node-only metastases (9.0 months) and lung-only metastases (8.1 months) had better survival compared to patients with liver-only metastases (4.6 months, P<0.001). Patients with multiple organ metastases had a median survival of 2.9 months which was significantly shorter than lung-only metastases, distant lymph node metastases-only, and liver-only metastases (all p<0.001).
However, after six months, the metastatic site lost its prognostic value with the median OS of all but multiple organ metastases not being significantly different. The number of patients who survived six months with liver-only metastases was 11,949 (21.1%). These patients with liver-only metastases had a median OS of 14.3 months compared to 15.6 months in patients with lung-only metastases (p=0.995) and distant lymph node-only metastases (15.9 months, p=0.136). The patients with multiple organ metastases had a median survival of 13.34 months which was significantly shorter than lung-only, distant lymph node-only, and liver-only metastases (Figure 2).
Figure 2.
(A) Kaplan-Meier curve of overall survival for patients at diagnosis. (B) Kaplan- Meier curve of overall survival for short survivors (patients who survived at least 6months) and (C) Kaplan-Meier curve of overall survival for long survivors (patients who survived at least 3 years)
In patients who survived beyond three years (N=2052, 3.6%), those with liver-only metastases (69.1 months) had a longer survival compared to those with lung-only metastases (56.8 months, P<0.001). The median survival for patients with liver-only metastases was also significantly longer than patients who had distant lymph node-only metastases (60.8 months, p=0.044).
These effects were also observed in the Aalen’s linear hazard model which indicated initial strong negative effect (shown as a steep incline of cumulative hazard for liver disease within the first 6 months), plateauing after six months (dissipating effect), and decline after 24 months (Figure 3a). Opposite effects were observed for lung-only metastatic site (Figure 3b) and non-time varying effects for peritoneal carcinomatosis (Figure 3c). Cramér-von Mises test showed significant change of effect between each progressive time cohorts for liver (p=0.018), and lung (p=0.019), but not for carcinomatosis (p=0.470) as the metastatic sites.
Figure 3.
(A) Aalen’s linear hazard model for patients with only liver metastases, (B) Aalen’s linear hazard model for patients with only lung metastases and (C)Aalen’s linear hazard model for patients with only carcinomatosis metastases
In the multivariate Cox regression analysis, older age, white, non-Hispanic and uninsured status were predictors of shorter OS from time of diagnosis. Having private insurance, Medicaid, or Medicare were associated with longer survival compared to not having insurance. Lung-only metastases, distant lymph node-only metastases, and bone-only metastases were associated with a statistically significant longer survival compared to liver-only metastases. On the other hand, other-only metastases and multiple organ metastases were associated with a statistically significant shorter survival. Finally, having had any chemotherapy, surgery of the primary site, or surgery of a distant site were all associated with longer survival. (Supplementary Material Table 2)
Factors associated with extended survival
In a logistic regression model predicting extended survival of three years or more, patients who were 65 years or older (OR = 0.632, 95% CI = 0.544 – 0.735, p<0.001) were less likely to survive for three or more years while females (OR = 1.18, 95% CI = 1.063 – 1.309, p=0.002) and Hispanic patients (OR=1.354, 95% CI=1.127 – 1.627, p=.001) were more likely to survive 3 or more years. Patients in the highest income quartile (≥ $74,063) were much more likely (OR = 1.415, 95% CI = 1.211 – 1.655, p<0.001) to reach the three-year survival compared to the lowest quartile (< $46,277).
Higher comorbidity was associated with lower odds (OR = 0.566, 95% CI = 0.433 – 0.741, p<0.001) of extended survival. Additionally, certain sites of metastasis, particularly carcinomatosis (OR = 0.165, 95% CI = 0.038 – 0.723, p=0.017) and multiple organ metastases (OR=.53, 95% CI=0.318 – 0.851, p=0.009) were associated with lower likelihood of extended survival. Finally, having surgery of the primary site (OR=7.858, 95% CI=6.651 – 9.284, p<0.001) and surgery of a distant site (OR=2.158, 95% CI=1.758 – 2.649, p<0.001) significantly increased the odds of surviving for at least 3 years after diagnosis (Table 2).
Table 2:
Multivariate logistic regression for survival ≥3 years after diagnosis
| OR | 95% CI | P value | |
|---|---|---|---|
|
| |||
| Age | |||
| <65 | Reference | ||
| ≥65 | 0.632 | (0.544 – 0.735) | <.001 |
| Sex | |||
| Male | Reference | ||
| Female | 1.18 | (1.063 – 1.309) | 0.002 |
| Race | |||
| White | Reference | ||
| Non-white | 1.071 | (0.977 – 1.175) | 0.143 |
| Ethnicity | |||
| Non-Spanish/Non-Hispanic | Reference | ||
| Hispanic | 1.354 | (1.127 – 1.627) | 0.001 |
| Primary Payor at Diagnosis | |||
| Not Insured | Reference | 0.002 | |
| Private Insurance | 1.232 | (0.907 – 1.674) | 0.181 |
| Medicaid | 0.98 | (0.696 – 1.381) | 0.909 |
| Medicare | 0.957 | (0.693 – 1.323) | 0.792 |
| Other Government | 0.633 | (0.341 – 1.176) | 0.148 |
| Median Income | |||
| < $46,277 | Reference | <.001 | |
| $46,277 – $57,856 | 0.919 | (0.766 – 1.104) | 0.367 |
| $57,857 – $74,062 | 1.083 | (0.911 – 1.287) | 0.368 |
| ≥$74,063 | 1.415 | (1.211 – 1.655) | <.001 |
| Charlson-Deyo Score | |||
| 0 | Reference | <.001 | |
| 1 | 0.879 | (0.771 – 1.001) | 0.053 |
| 2 | 0.583 | (0.452 – 0.753) | <.001 |
| ≥ 3 | 0.566 | (0.433 – 0.741) | <.001 |
| Mets at Diagnosis - Multiple Organs | |||
| No | Reference | ||
| Yes | 0.52 | (0.318 – 0.851) | 0.009 |
| Mets at Diagnosis - Only Other (other than bone, brain, liver, lung or distant lymph nodes) | |||
| No | Reference | ||
| Yes | 0.671 | (0.403 – 1.115) | 0.123 |
| Mets at Diagnosis - Only Carcinomatosis | |||
| No | Reference | ||
| Yes | 0.165 | (0.038 – 0.723) | 0.017 |
| Mets at Diagnosis - Only Distant Lymph Node | |||
| No | Reference | ||
| Yes | 0.975 | (0.539 – 1.764) | 0.934 |
| Mets at Diagnosis - Only Liver | |||
| No | Reference | ||
| Yes | 0.84 | (0.518 – 1.364) | 0.481 |
| Mets at Diagnosis – Only Lung | |||
| No | Reference | ||
| Yes | 1.015 | (0.602 – 1.713) | 0.954 |
| Any Surgery | |||
| No Surgery | Reference | ||
| Any Surgery of Primary Site | 7.858 | (6.651 – 9.284) | <.001 |
| Any Surgery at Distant Site | |||
| No Surgery | Reference | ||
| Any Surgery of Distant Site | 2.158 | (1.758 – 2.649) | <.001 |
| Any Chemotherapy | |||
| No Chemotherapy | Reference | ||
| Yes, received chemotherapy | 1.046 | (0.938 – 1.166) | 0.42 |
In a sub analysis of the 15,661 (28.9%) patients who had reported values for their CA19–9 levels, those patients who survived greater than three years after diagnosis had an average CA19–9 value of 576.4 (SD 416.4) which was significantly lower than the average CA19–9 value of 784.5 (SD 352.2) for patients who survived less than three years after their initial diagnosis (p<.001).
DISCUSSION
In this study we determined the role of the site of metastasis at diagnosis on survival over time and also assessed which factors were associated with extended survival in patients with metastatic PDAC. This study showed, as other retrospective studies have, that liver metastases at diagnosis predicts worse survival compared to other sites of metastases.27 This is in line with the findings of the MPACT (Metastatic Pancreatic Adenocarcinoma Clinical Trial) trial which found that liver metastases were associated with worse overall survival as well as a retrospective study done by He et al. which had similar findings.18, 27 In another study, Oweira et al. looked at the metastatic sites at time of diagnosis of PDAC and found that patients with PDAC who had isolated liver metastases had worse outcomes compared to patients with isolated lung or distant nodal metastases.19 While our study confirmed these findings, we found that this prognostic value was lost once a patient survived at least six months after diagnosis. At six months of survival, there was no difference in overall survival for patients with isolated liver metastases compared to all other sites of metastasis besides multiple organs. To our knowledge, this is the first time that the prognostic value of sites of metastases at diagnosis has been evaluated longitudinally at different time points.
These findings may be due to a multitude of factors affecting the survival of metastatic PDAC patients. First, patients with fast progressive and exceedingly unfavorable disease as attributed to liver metastatic disease are negatively selected faster within the first six months.13, 21 However, beyond six months and therefore approximately the length of induction treatment when a curative approach is chosen, the site of metastasis loses its prognostic value. After that negative selection, patients with oligometastatic disease may profit from surgical resection regardless of metastatic site except for multiple sites of metasatses.14, 28–30 Further, factors that influence this negative selection may be higher tumor burden, no treatment or very poor functional status at diagnosis, thus suggesting a need for biological and conditional selection of patients for further intervention.31
The current first-line treatment for metastatic pancreatic cancer is induction therapy with FOLFIRINOX32. This treatment, however, has its limitations as it is very toxic resulting in side effects that are often hard for patients to tolerate.33 Because of this, the mode of treatment has to be chosen carefully. However, in those willing to undergo aggressive treatment with metastatic disease, surgery might be an acceptable option improving prognosis once they have passed the so called biological litmus test.14, 29, 30 Although the evidence for surgical resection in PDAC is sparse and its role is highly disputed, survival benefit of this approach has been shown in certain types of cancer such as colorectal and gastric cancer.34, 35 Some studies have shown a benefit of resection in metastatic PDAC such as a study of 64 patients with liver metastasis that showed patients who underwent resection of both the primary and metastatic site had significantly longer overall survival.36 Our study showed that having surgery at the primary tumor site and or at the distant metastatic site was associated with longer overall survival suggesting a possible benefit in a subset of patients. There are few studies that have analyzed long-term survival PDAC.25, 37 Extended survivors in metastatic PDAC have been defined with different cut-offs in the literature. Studies by Goulart et al. and Inal et al. defined “long survivors” as greater than 1 year while another study chose a longer cutoff for extended survival of 5 years.23, 24, 37 Similar to our study, that defined extended survival as surviving at least three years after diagnosis, Jo et al. found better performance status was related to longer survival but impaired survival was seen in patients with peritoneal metastasis, and lower CA19–9.25 Our study also showed a correlation between lower CA19–9 levels and extended survival of greater than 3 years.
Performance status and comorbidities have been shown to be an important predictor for tolerability of cancer directed treatment and subsequently extended survival and better outcomes in patients with pancreatic cancer.38 Sezgin et al. found that in patients with advanced pancreatic carcinoma who were treated with gemcitabine, those with low performance status as defined by ECOG had worse survival.39 In our study, performance status was evaluated using the Charlson-Deyo index, and we found that higher scores were associated with lower odds of achieving extended survival of three years. This finding shows that the overall health and performance status of a patient is an integral factor in determining treatment and predicting the survival outcomes of patients with metastatic PDAC. Our study indicates that while aggressive treatment is associated with extended survival, it must be well-tolerated by the patient, highlighting the importance of considering performance status and comorbidities when determining the treatment approach.
Outcomes of pancreatic cancer are associated with multiple factors involving demographics and socioeconomic status. A previous study by Thomas et al. showed that patients with private insurance are more likely to have access to high volume cancer centers.40 Many studies also have shown that insurance status plays a role in whether patients receive cancer directed therapies.41–43Insurance status has also been shown to be a negative predictor of surgical resection in patients with PDAC.44, 45 Our findings were in line with the literature as we found that patients with private insurance, Medicaid or Medicare were all associated with increased odds of surviving for three or more years (extended survivors) compared to those patients who were uninsured. Our analysis also found that patients in the highest income quartile had significantly higher odds of being an extended survivor.
Strengths and Limitations
This is the first study that reports that the prognostic value of metastatic sites changes over time. The current study utilized two different methods to do this, using both Kaplan Meier Curves as previous studies have and Aalen’s linear hazard models which is novel. This study also adds evidence of predictors of extended survival in metastatic PDAC.
However, the current analysis does have many limitations in addition to those discussed earlier regarding selection bias, as it is a retrospective analysis that was conducted using a very large database. The NCDB has many advantages, but it also lacks granularity. While we are able to tell if a patient had received chemotherapy, it is not possible to determine the specific therapies used or how many cycles each patient received. The NCDB is also made up of reporting from different hospitals and facilities from across the United States. This leads to inconsistencies in reporting even with stringent guidelines associated with the NCDB reporting.
Another limitation was that within our logistic regression model there was a class imbalance where only 3.6% of the study population were survivors at the three-year mark. However, this discrepancy did not significantly impact the performance of the model, as our receiver operating characteristic area, a widely used measure of classification performance, achieved a value of 0.7.
Within the cox regression model, we saw evidence suggesting a departure from the proportional hazards’ assumption with the Schoenfeld residuals proportional hazards test being significant (p<.001) as also stated by the Aalen linear model for our primary outcome. By utilizing both time-averaged effects and the Aalen linear hazards model, we ensured a comprehensive examination of the data, mitigating the impact of non-proportional hazards and reinforcing the validity of our conclusions.
In conclusion, while at diagnosis liver-only metastases predicts worse outcomes for patients with metastatic PDAC as compared to lung-only and lymph node-only metastases, this predictive value is lost once patients make it to six months of survival. Therefore, beyond this window, treatment decisions should not primarily be based on the site of metastatic disease. Extended survival may be achieved in a small subset of patients. This subset of patients is defined by surrogate markers of more favorable tumor biology and better conditional status making them more likely to undergo aggressive treatment including primary tumor resection and achieve favorable survival.
Supplementary Material
Footnotes
Conflicts of interests: There are no conflicts of interest for any of the authors.
Conflicts of Interest:
None declared.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Society AC. Cancer Facts & Figures 2023. Journal [serial online]. 2023. Available from: American Cancer Society. [Google Scholar]
- 2.Ghabi EM, Shoucair S, Ding D, et al. Tailoring Adjuvant Chemotherapy to Biologic Response Following Neoadjuvant Chemotherapy Impacts Overall Survival in Pancreatic Cancer. J Gastrointest Surg 2023; 27(4):691–700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Javed AA, Ding D, Hasanain A, et al. Persistent Circulating Tumor Cells at One Year after Oncologic Resection Predict Late Recurrence in Pancreatic Cancer. Ann Surg 2022. [DOI] [PubMed] [Google Scholar]
- 4.Kinny-Köster B, Habib JR, Wolfgang CL, et al. Favorable tumor biology in locally advanced pancreatic cancer-beyond CA19–9. J Gastrointest Oncol 2021; 12(5):2484–2494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Chen X, Liu F, Xue Q, et al. Metastatic pancreatic cancer: Mechanisms and detection (Review). Oncol Rep 2021; 46(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Rhim Andrew D, Mirek Emily T, Aiello Nicole M, et al. EMT and Dissemination Precede Pancreatic Tumor Formation. Cell 2012; 148(1):349–361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Adamska A, Domenichini A, Falasca M. Pancreatic Ductal Adenocarcinoma: Current and Evolving Therapies. Int J Mol Sci 2017; 18(7). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sohal DPS, Kennedy EB, Cinar P, et al. Metastatic Pancreatic Cancer: ASCO Guideline Update. Journal of Clinical Oncology 2020; 38(27):3217–3230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pedrazzol S. Currently Debated Topics on Surgical Treatment of Pancreatic Ductal Adenocarcinoma: A Narrative Review on Surgical Treatment of Borderline Resectable, Locally Advanced, and Synchronous or Metachronous Oligometastatic Tumor. Journal of Clinical Medicine 2023; 12(20):6461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chen Y, Che X, Zhang J, et al. Long-term results of intraoperative electron beam radiation therapy for nonmetastatic locally advanced pancreatic cancer: Retrospective cohort study, 7-year experience with 247 patients at the National Cancer Center in China. Medicine (Baltimore) 2016; 95(38):e4861. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Steen W, Blom R, Busch O, et al. Prognostic value of occult tumor cells obtained by peritoneal lavage in patients with resectable pancreatic cancer and no ascites: A systematic review. J Surg Oncol 2016; 114(6):743–751. [DOI] [PubMed] [Google Scholar]
- 12.Epstein JD, Kozak G, Fong ZV, et al. Microscopic lymphovascular invasion is an independent predictor of survival in resected pancreatic ductal adenocarcinoma. J Surg Oncol 2017; 116(6):658–664. [DOI] [PubMed] [Google Scholar]
- 13.Habib JR, Kinny-Köster B, Bou-Samra P, et al. Surgical Decision-Making in Pancreatic Ductal Adenocarcinoma: Modeling Prognosis Following Pancreatectomy in the Era of Induction and Neoadjuvant Chemotherapy. Ann Surg 2023; 277(1):151–158. [DOI] [PubMed] [Google Scholar]
- 14.Ilmer M, Schiergens TS, Renz BW, et al. Oligometastatic pulmonary metastasis in pancreatic cancer patients: Safety and outcome of resection. Surg Oncol 2019; 31:16–21. [DOI] [PubMed] [Google Scholar]
- 15.Deeb A, Haque SU, Olowokure O. Pulmonary metastases in pancreatic cancer, is there a survival influence? J Gastrointest Oncol 2015; 6(3):E48–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Liu Z, Gou A, Wu X. Liver metastasis of pancreatic cancer: the new choice at the crossroads. Hepatobiliary Surg Nutr 2023; 12(1):88–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Embuscado EE, Laheru D, Ricci F, et al. Immortalizing the complexity of cancer metastasis: genetic features of lethal metastatic pancreatic cancer obtained from rapid autopsy. Cancer Biol Ther 2005; 4(5):548–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.He C, Huang X, Zhang Y, et al. The impact of different metastatic patterns on survival in patients with pancreatic cancer. Pancreatology 2021; 21(3):556–563. [DOI] [PubMed] [Google Scholar]
- 19.Oweira H, Petrausch U, Helbling D, et al. Prognostic value of site-specific metastases in pancreatic adenocarcinoma: A Surveillance Epidemiology and End Results database analysis. World J Gastroenterol 2017; 23(10):1872–1880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Kruger SF, Lohneis A, Abendroth A, et al. Prognosis and tumor biology of pancreatic cancer patients with isolated lung metastases: translational results from the German multicenter AIO-YMO-PAK-0515 study. ESMO Open 2022; 7(1):100388. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Rompen IF, Levine J, Habib JR, et al. Progression of Site-Specific Recurrence of Pancreatic Cancer and Implications for Treatment. Ann Surg 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhang L, Jin R, Yang X, et al. A population-based study of synchronous distant metastases and prognosis in patients with PDAC at initial diagnosis. Front Oncol 2023; 13:1087700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Goulart BHL, Clark JW, Lauwers GY, et al. Long term survivors with metastatic pancreatic adenocarcinoma treated with gemcitabine: a retrospective analysis. Journal of Hematology & Oncology 2009; 2(1):13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Inal A, Ciltas A, Yildiz R, et al. Long term survivors with metastatic pancreatic cancer treated with gemcitabine alone or plus cisplatin: a retrospective analysis of an Anatolian Society of Medical Oncology multicenter study. Asian Pac J Cancer Prev 2012; 13(5):1841–4. [DOI] [PubMed] [Google Scholar]
- 25.Jo JH, Chung MJ, Park JY, et al. Clinical Characteristics of Long-Term Survivors of Inoperable Pancreatic Cancer: An 8-Year Cohort Analysis in Korea. Pancreas 2014; 43(7). [DOI] [PubMed] [Google Scholar]
- 26.Hosmer DWR P. Using Aalen’s linear hazards model to investigate time-varying effects in the proportional hazards regression model. Stata Journal 2002; 2(4):20. [Google Scholar]
- 27.Tabernero J, Chiorean EG, Infante JR, et al. Prognostic factors of survival in a randomized phase III trial (MPACT) of weekly nab-paclitaxel plus gemcitabine versus gemcitabine alone in patients with metastatic pancreatic cancer. Oncologist 2015; 20(2):143–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Dünschede F, Will L, von Langsdorf C, et al. Treatment of metachronous and simultaneous liver metastases of pancreatic cancer. Eur Surg Res 2010; 44(3–4):209–13. [DOI] [PubMed] [Google Scholar]
- 29.Hackert T, Niesen W, Hinz U, et al. Radical surgery of oligometastatic pancreatic cancer. European Journal of Surgical Oncology (EJSO) 2017; 43(2):358–363. [DOI] [PubMed] [Google Scholar]
- 30.Homma Y, Endo I, Matsuyama R, et al. Outcomes of lung metastasis from pancreatic cancer: A nationwide multicenter analysis. Journal of Hepato-Biliary-Pancreatic Sciences 2022; 29(5):552–561. [DOI] [PubMed] [Google Scholar]
- 31.Michelakos T, Pergolini I, Castillo CF-d, et al. Predictors of Resectability and Survival in Patients With Borderline and Locally Advanced Pancreatic Cancer who Underwent Neoadjuvant Treatment With FOLFIRINOX. Annals of Surgery 2019; 269(4). [DOI] [PubMed] [Google Scholar]
- 32.Vaccaro V, Sperduti I, Milella M. FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer. N Engl J Med 2011; 365(8):768–9; author reply 769. [DOI] [PubMed] [Google Scholar]
- 33.Gunturu KS, Yao X, Cong X, et al. FOLFIRINOX for locally advanced and metastatic pancreatic cancer: single institution retrospective review of efficacy and toxicity. Medical Oncology 2012; 30(1):361. [DOI] [PubMed] [Google Scholar]
- 34.McNally SJ, Parks RW. Surgery for colorectal liver metastases. Dig Surg 2013; 30(4–6):337–47. [DOI] [PubMed] [Google Scholar]
- 35.Takemura N, Saiura A, Koga R, et al. Long-term outcomes after surgical resection for gastric cancer liver metastasis: an analysis of 64 macroscopically complete resections. Langenbecks Arch Surg 2012; 397(6):951–7. [DOI] [PubMed] [Google Scholar]
- 36.Tachezy M, Gebauer F, Janot M, et al. Synchronous resections of hepatic oligometastatic pancreatic cancer: Disputing a principle in a time of safe pancreatic operations in a retrospective multicenter analysis. Surgery 2016; 160(1):136–144. [DOI] [PubMed] [Google Scholar]
- 37.Kardosh A, Lichtensztajn DY, Gubens MA, et al. Long-Term Survivors of Pancreatic Cancer: A California Population-Based Study. Pancreas 2018; 47(8). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Oba A, Croce C, Hosokawa P, et al. Prognosis Based Definition of Resectability in Pancreatic Cancer: A Road Map to New Guidelines. Annals of Surgery 2022; 275(1). [DOI] [PubMed] [Google Scholar]
- 39.Sezgin C, Karabulut B, Uslu R, et al. Gemcitabine treatment in patients with inoperable locally advanced/metastatic pancreatic cancer and prognostic factors. Scand J Gastroenterol 2005; 40(12):1486–92. [DOI] [PubMed] [Google Scholar]
- 40.Thomas AS, Sharma RK, Kwon W, et al. Socioeconomic Predictors of Access to Care for Patients with Operatively Managed Pancreatic Cancer in New York State. Journal of Gastrointestinal Surgery 2022; 26(8):1647–1662. [DOI] [PubMed] [Google Scholar]
- 41.Bilimoria KY, Bentrem DJ, Ko CY, et al. National failure to operate on early stage pancreatic cancer. Ann Surg 2007; 246(2):173–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Khanal N, Upadhyay S, Dahal S, et al. Systemic therapy in stage IV pancreatic cancer: a population-based analysis using the National Cancer Data Base. Ther Adv Med Oncol 2015; 7(4):198–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Wright MJ, Overton HN, Teinor JA, et al. Disparities in the Use of Chemotherapy in Patients with Resected Pancreatic Ductal Adenocarcinoma. J Gastrointest Surg 2020; 24(7):1590–1596. [DOI] [PubMed] [Google Scholar]
- 44.Molina G, Clancy TE, Tsai TC, et al. Racial Disparity in Pancreatoduodenectomy for Borderline Resectable Pancreatic Adenocarcinoma. Annals of Surgical Oncology 2021; 28(2):1088–1096. [DOI] [PubMed] [Google Scholar]
- 45.Moaven O, Richman JS, Reddy S, et al. Healthcare disparities in outcomes of patients with resectable pancreatic cancer. Am J Surg 2019; 217(4):725–731. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.