Abstract
Pancreatic adenocarcinoma, increasingly diagnosed in the United States, has a disheartening initial resection rate of 15%. Neoadjuvant therapy, particularly FOLFIRINOX and gemcitabine-based regimens, is gaining favor for its potential to improve resectability rates and achieving microscopically negative margins (R0) in borderline resectable cases, marked by intricate arterial or venous involvement. Despite surgery being the sole curative approach, actual benefit of neoadjuvant therapy remains debatable. This study scrutinizes current literature on oncological outcomes post-resection of borderline resectable pancreatic cancer. A MEDLINE/PubMed search was conducted to systematically compare oncological outcomes of patients treated with either neoadjuvant therapy with intent of curative resection or an “upfront resection” approach. A total of 1293 studies were initially screened and 30 were included (n = 1714) in this analysis. All studies included data on outcomes of patients with borderline resectable pancreatic adenocarcinoma being treated with neoadjuvant therapy (n = 1387) or a resection-first approach (n = 356). Patients treated with neoadjuvant therapy underwent resection 52% of the time, achieving negative margins of 43% (n = 601). Approximately 77% of patients who received an upfront resection underwent a successful resection, with 39% achieving negative margins. Neoadjuvant therapy remains marginally efficacious in treatment of borderline resectable pancreatic adenocarcinoma, as patients undergo an operation and successful resection less often when treated with neoadjuvant therapy. Rates of curative resection are comparable, despite neoadjuvant therapy being a primary recommendation in borderline resectable cases and employed more often than upfront resection. Upfront resection may offer improved resection rates by intention-to-treat, which can provide more patients with paths to curative resection.
Introduction
Pancreatic cancer remains to rise in occurrence eminently and problematically, and its prognosis poor. In the United States, the incidence of pancreatic cancer is estimated to increase by 64 000 in 2023, challenging many with a dismal 5-year survival rate of only 12%. 1 With it being the third leading cause of cancer-related death in America, a spot previously held by colorectal cancer just years prior, the only curative solution for patients is complete resection. Margin status is a significant risk factor for overall survival. 2 Microscopically negative margins (R0 resection) are associated with a 5-year survival of 20%3,4 with median overall survival between 17 and 23 months.3-5 The desire for improved oncological and survival outcomes for patients with borderline resectable pancreatic adenocarcinoma has greatly motivated research into neoadjuvant care due to the complex nature and question of resectability in this diagnosis.
Borderline resectable pancreatic adenocarcinoma is a preoperative diagnosis of pancreatic adenocarcinoma which is a high risk for margin-positive resection. It was initially defined in 2005 by Varadhachary et al of MD Anderson, and further by an international consensus statement by the Americas Hepato-Pancreato-Biliary Association, Society of Surgical Oncology and Society for Surgery and the Alimentary Tract (AHPBA/SSO/SSA) in 2009.6,7 The current National Comprehensive Cancer Network (NCCN) also continually updates its definition for treatment recommendations in its annual guidelines. Many definitions have consolidated from purely anatomical to a recent consensus statement by the International Association of Pancreatology (IAP) in 2017 emphasizing biological, conditional, and anatomical factors such as carbohydrate antigen 19-9 markers (CA 19-9). 8 The major point of contentment between these definitions relies on the extent of vascular involvement and more recently non-anatomical signatures of this diagnosis.6-9 This is an important diagnosis to ensure consistency, as patients with borderline resectable pancreatic cancer sit at the precipice of decisions and outcomes.
In the last 20 years, neoadjuvant therapy has become a hotspot for research into standardized treatments for patients with pancreatic adenocarcinoma. Preoperative chemotherapy was proposed to achieve improved resectability of unresectable, locally advanced, and borderline resectable staging. Gemcitabine was demonstrated to be comparable with 5-fluorouracil in achieving similar median survival with a mild advantage over 5-fluorouracil (5.7-month and 4.4-month median survival respectively). 10 Meta-analysis indicated that preoperative gemcitabine is safe and efficacious for patients with resectable and unresectable pancreatic cancer, resulting in an 89% R0 resection rate, 11 which raised the question of possible gemcitabine combination regimens. A phase II trial comparing gemcitabine/cisplatin with upfront resection showed comparable results between the 2 strategies, with 16/33 (48%) patients and 23/33 (70%) patients undergoing resection respectively, but R0 resection rates were found to be 89% (17/19) and 69% (16/23), respectively. 12 Combination of fluorouracil, irinotecan, and oxaliplatin (FOLFIRINOX) has been recommended with evidence showing a 10-month superior survival outcome to gemcitabine/nab-paclitaxel (38.7 months vs 28.6 months). 13 In one study, neoadjuvant therapy was shown to be associated with 80% decline of CA 19-9, contributing to the enthusiasm behind preoperative chemotherapy on even novel methods used to inform pancreatic cancer diagnosis. 14
Neoadjuvant therapy has been shown to be safe and effective, yet there are limited review data focused on intention-to-treat numbers on survival to surgery, resection, and R0 resection rates as well as a limited description of disease progression in patients with borderline resectable pancreatic cancer during neoadjuvant therapy, particularly when comparing to upfront resection. In 2013, Toomey et al concluded that there is a severe reduction in patients undergoing resection after neoadjuvant therapy compared to the resection-first approach. Overall, patients lived more months total and per patient in upfront resection cohorts compared to neoadjuvant counterparts (1553 months, 53% overall R0 resection rate and 1354 months, 42% overall R0 resection rate respectively). 15 10 years later, we are revisiting this endeavor to examine the literature and evaluate if progress has been made with the advent of new neoadjuvant treatments and more rigorous clinical trials.
This study was undertaken to systematically review the literature on neoadjuvant therapy for patients with borderline resectable pancreatic adenocarcinoma and the associated outcomes. We intend to focus on intention-to-treat resection rates as indicators of long-term survival and curability. Additionally, we aim to identify the complications occurring during neoadjuvant therapy that result in the suspension of treatment, inoperability, and unresectability. This will provide a clearer understanding of the neoadjuvant therapy platform. We hypothesize that patients receiving neoadjuvant therapy will have improved rates of resection and a comparable or greater than R0 resection rate by intention-to-treat compared to patients receiving upfront resection.
Methods
A MEDLINE/PubMed search was systematically conducted and concluded in 2023 using keywords “borderline resectable pancreatic adenocarcinoma,” “borderline resectable pancreatic AND chemo*” as well as “borderline resectable pancreatic AND neoadjuvant.” This yielded 1293 results from 2005 through 2023. The selection was limited to these dates due to the adoption of standardized definitions of borderline resectable pancreatic as previously described.6-9 A manual addition of 4 studies was included to increase statistical power and due to their relevance to the hypothesis.
Qualitative analysis was initiated and conducted by three reviewers J.P., M.C., and T.P working non-independently and reviewed all 85 papers. Qualitative assessment was made up of a select set of questions as well as based on specific exclusion parameters. Studies were required to demonstrate sufficient relevancy to the research question in hand by answering undoubtedly their own research question and ours, as studies with novel or poorly generalizable treatment regimens were excluded. Studies were further excluded if they failed to demonstrate isolation of groups (borderline, locally advanced, unresectable, etc.) when reporting outcomes and endpoints, and were required to report intention-to-treat resection rates or otherwise convertible measures and margin status. Papers that failed to make transparent their definitions of borderline resectable pancreatic cancer, inclusion criterion, and their treatment regimens were additionally excluded.
All papers were compiled together and results separated into 2 groups: patients treated with neoadjuvant therapy and patients treated with an upfront resection approach. Data was abstracted following qualitative analysis. Outcomes which were sought out included resection rate (patients undergoing resection/patients undergoing an operation), R0 resection rate (patients achieving R0 margins/patients undergoing resection), intention-to-treat (ITT) R0 resection rate (patients achieving R0 margins/total sample size), margin status (R0, R1, R2, and Rx) and overall survival. Papers with missing data of operations, resection or margin status were excluded unless sample size and all patient outcomes were provided such that these values could be calculated. All other variables gathered included the study design style, institution and location, radiotherapy and chemotherapy administered, complications and reasons for suspension of neoadjuvant therapy, and overall survival. Data abstracted were stored and figures compiled on secure Microsoft© Office 16 Software. This literature review included no assessments for homogeneity, risk of bias, sensitivity assessment, or certainty between studies included. All combined values between studies are done so with the acknowledgment of potential heterogeneity given the wide scope of the studies’ treatments and measured outcomes.
Results
A total of 1293 studies were included in initial screening following our literature search, followed by exclusion and qualitative analysis. A total of 326 results were excluded due to being review articles, as well as 413 articles which were opinion pieces (84), consensus statements (67), case studies (43), protocols/technique descriptions (165), or guidelines (54). Following the title and abstract analysis, further studies not relevant or pertinent to the topic of neoadjuvant chemotherapy in patients with strictly borderline resectable pancreatic adenocarcinoma undergoing surgery were excluded (469) A total of 4 studies were manually added upon citation searching in other papers, which is a potential risk for screening bias. The remaining group of 85 studies were available for full-text qualitative analysis. An example of studies removed includes 7 studies removed for poor generalizability to the United States population given their usage of S-1 gemcitabine regimens, which require the S-1 genotype in patients selected for this therapy16-18 [Figure 1]. Our systematic review yielded 30 studies in which patient cohorts with borderline resectable pancreatic adenocarcinoma were treated with neoadjuvant therapy and whose surgical and oncological outcomes could be isolated and analyzed9,19-47 [Table1]. Twenty-one studies only examined neoadjuvant therapy prior to resection, one study included patients treated with an upfront resection approach, and nine studies included both cohorts of patients who underwent neoadjuvant therapy or upfront resection. Fourteen out of thirty of these studies were prospective designs. Neoadjuvant first-line chemotherapy regimens included Gemcitabine, FOLFIRINOX, Cisplatin, and combination regimens including GEMOX, Gemcitabine + nab-paclitaxel, docetaxel/capecitabine, and several others [Table 1]. Flowchart of paper selection process from identification, screening, eligibility, and inclusion. *Relevancy was reassessed following further exclusion parameters upon consensus that the pool of studies was not sufficiently specific and/or did not address the topic of outcomes of patients with borderline pancreatic adenocarcinoma treated with neoadjuvant therapy and then surgical resection. Re-inclusion parameters included “neoadjuvant chemo* AND resection* AND borderline resectable”. Studies That Include Patients Who Underwent Neoadjuvant Therapy for Borderline Resectable Pancreatic Adenocarcinoma. 5FU (5-Fluorouracil); Cis (Cisplatin); Leuc (Leucovorin); Doc (Docetaxel); Cap (Capecitabine); Intensity-Modulated Radiation Therapy (IMRT); Computer-Controlled Radiation Therapy (CCRT), Chemoradiotherapy (CRT), Stereotactic-Body Radiotherapy (SBRT), Three-Dimensional Radiotherapy (3DRT), External Beam Radiotherapy (EBRT), Radiation Therapy (RT).
Patients With Borderline Resectable Pancreatic Adenocarcinoma Who Underwent Upfront Resection (n = 9 Studies).
Patients Who Underwent an Operation After Neoadjuvant Therapy; Tran Includes 5 Patients Not Finish Neoadjuvant Therapy in Final Analysis; Maggino Includes 4 Patients Not Finish Neoadjuvant Therapy in Final Analysis; R0 (Microscopically/Macroscopically Negative).
Flowcharts of outcomes for patients who underwent neoadjuvant therapy. All studies (n = 29) with intent-to-treat operative outcome data on patients treated with neoadjuvant therapy in an operated vs non-operated pathway chart with 43% of patients achieving R0 resection (A). All studies (n = 8) with intent-to-treat operative outcome data on patients treated with an upfront resection approach in a similar pathway chart with 39% of patients achieving R0 resection (B).
A total of 19 studies (n = 1087) reported patients having suspended or interrupted their preoperative chemotherapy in 176 cases (16%). Nine papers excluded patients who did not finish neoadjuvant therapy from further analysis, or specified which patients underwent an operation or not. Of 526 patients, 424 (81%) completed neoadjuvant therapy. The negative margin resection rate by intention-to-treat for studies that eliminated failure of neoadjuvant therapy from further cohorts was 32%. Reasons for not proceeding with the operation following neoadjuvant therapy included death in 16 cases, disease progression in 105 cases (36 were found with metastatic disease, 19 with locally advanced and 12 with unresectable disease before the operation), 13 cases with comorbidity conflict, 11 cases where failure of neoadjuvant therapy excluded from the operation and a total of 342 patients had unspecified ineligibility for the operation. 123 patients in the neoadjuvant therapy arm had to abort resection at the time of operation due to having locally advanced disease (10), metastatic progression (17), becoming surgically unresectable (11), or otherwise unspecified abortions (79). In the upfront resection arm, 47 patients became ineligible for surgery between the time of inclusion and surgery due to death (3), comorbidity (1), refusal (5), or generalized unspecified ineligibility (38). 56 patients had aborted resections at the time of operation due to unspecified disease progression or other complications.
A total of 27 studies included any survival data per treatment group, and only 16 studies included isolated median overall survival data for borderline resectable pancreatic adenocarcinoma, contributing to the heterogeneity between study treatments and reported outcomes. Nine of which admitted to using survival as a primary endpoint [Figure 3]. The range of median overall survival in the neoadjuvant arm is between 12 months and 35.4 months for patients diagnosed with borderline resectable tumors. Data was severely limited in upfront resection arm. In one study Kang et al, the upfront resection arm included patients who were diagnosed with resectable pancreatic cancer and found that the results of patients with resectable cancers undergoing no neoadjuvant therapy before an operation were comparable with the neoadjuvant arm (P > .06).
27
Across the board, successful resections and neoadjuvant therapy studies were associated with higher median survivals, and overall higher percent survived at the 1-, 2-, and 5-year marks [Table 4]. Flowchart of 100 patients who theoretically undergo neoadjuvant therapy prior to resection and 100 patients who theoretically undergo an upfront operation for borderline resectable pancreatic adenocarcinoma. Survival data Abstracted From all 30 Studies. Median Overall Survival (OS), BRPC (Borderline Resectable Pancreatic adenocarcinoma), LAPC (Locally Advanced Pancreatic Adenocarcinoma) Not yet Reached (NYR), Non-resected (-Rx), Resected (+Rx), Adjuvant Therapy (+Adj), No Adjuvant Therapy (-Adj), Median Follow-up Time (mFT), 1-/2-/5-year Overall Survival (1Y, 2Y, and 5Y). aValues represent combined cohorts of borderline resectable with LAPC, resectable and/or unresectable. bNo treatment or surgeries were conducted in this group. cSurvival was calculated at time of enrollment, not time of surgery.
According to medical literature, the median survival of patients who did not undergo a resection for pancreatic adenocarcinoma is estimated to be 8 months; the estimated median survival of patients who underwent either a positive margin resection (R1/R2) or negative margin resection (R0) is approximately 10 months and 21 months, respectively.48-50 Given this, the calculated total months survived per patient treated with neoadjuvant in a theoretical model, out of all 30 studies (n = 1743), is 17.7 months, with patients undergoing a resection-first approach with intent for curative resection having an estimated 15.8 months survived per patient. Out of 356 patients in upfront resection groups, 15% became inoperable or unresectable. During treatment with neoadjuvant therapy, 36% became inoperable or unresectable [Figure 2].
Discussion
Borderline resectable pancreatic cancer remains to be not only a dismal prognosis, but one with unclear paths to curative resection. It is well understood that resection is the only option to cure borderline resectable pancreatic adenocarcinoma. Neoadjuvant therapy has been offered as the solution, with many referencing the high R0 resection rates—perhaps one of the best indicators for long-term survival and curability, particularly when compared to upfront resection. 2 Improving odds of microscopically negative margins would greatly benefit the patients sitting at the tenuous stage of borderline resectability, as it is well-discussed that neoadjuvant therapy potentially offers the ability to downstage tumors, increasing resectability in particularly complex cases involving arterial and venous involvement. There has been much enthusiasm behind neoadjuvant therapy, cited as even addressing metastatic lymph node progression.9,49,51 There is a need to take a critical look at the metrics we are using to determine curability, given the heterogeneity of available survival data, dichotomous data such as margin and resection status offer more consistent datapoints, until standardized and transparent methods of listing all cohort median survival and long-term survival data are made available.
Studies examining outcomes of neoadjuvant therapy are often retrospective, which often fail to report intention-to-treat numbers that leave out crucial whole sample values. Traditionally in literature, resection rate and R0 resection rate are denominator-reducing endpoints, in which the denominator used is smaller than the true sample of patients who are intended to be treated. This can inflate measures of success. If fifty patients are intended to be treated with chemotherapy, and 30 of them undergo a resection with 25 having an R0 resection, then the traditional R0 rate would be 83% whereas the intention-to-treat R0 resection rate is 50%. This review found that on average, studies using limiting denominators have eliminated 35% of their cohort from analysis when using resection rate reported as successful resections over patients operated. As much as 42% of patients on average are eliminated from analysis when using R0 resection rate reported as the amount of R0 resections over successful resection. Conducting intention-to-treat-based prospective trials helps curb this problem, as more patients can be identified in a real-world perspective, in lieu of utilizing data from compliant subsamples to a protocol. The theoretical model generated is not intended to serve as a meta-analysis true estimate, but simply a guideline for how the medical literature should understand outcomes in pancreatic cancer, since reframing the outcomes to properly reflect real-world clinic scenarios, with failures, ineligibility, toxicity, open-close surgeries, and positive margins, will better represent the data on neoadjuvant therapy.
Studies examining borderline resectable cohorts often include other groups such as resectable or locally advanced. Inclusion of these groups which sit at opposite ends of potential outcomes tips the scales of the data. Patients either have resectable diseases already fit for an operation or patients with locally advanced and metastatic disease have low probability for turnover and curative resection. It was critical during this analysis to ensure all papers isolated outcome—both oncological and survival—data of patients with borderline resectable pancreatic cancer. During qualitative analysis, 41 papers failed to isolate patients with borderline resectable adenocarcinoma when reporting outcome data. Further incertitude results from the evolution of definitions of borderline resectable, hinging on the interface between the tumor and vascular wall. Most recently there has been the introduction of utilizing biological factors such as CA 19-9 levels as an estimation for determining resectability. A consensus statement by the International Association of Pancreatology notes the significance of CA 19-9 levels >500u/mL as suspicion for, but not evidence for, distant metastases and therefore an indication for borderline resectable, 8 however elevated tumor marker CA 19-9 could indicate anything from malignant obstruction to cholangitis as emphasized by NCCN guidelines. 52 This could contribute to the incertitude as it continues to further estimations surrounding this diagnosis.
Limitations of this paper include that no meta-analysis was performed when generating the theoretical model of determining outcomes utilizing a months-per-patient lived method. This literature review included no assessments for homogeneity, risk of bias, sensitivity assessment, or certainty between studies. All combined values between studies were considered with the acknowledgment of potential heterogeneity, representing a theoretical predictive model that requires further validation. The approach of assessing outcomes is based on modeling real-world clinical scenarios, reflecting a variety of possible outcomes that a patient might face when presenting for treatment. This method is useful for contributing to the medical literature on neoadjuvant therapy, as it aims to represent data on neoadjuvant therapy more accurately by considering failures, ineligibility, toxicity, positive margins, and other factors. Furthermore, there is a potential risk of bias in study selection due to the use of mostly automated exclusion methods prior to qualitative analysis. The emphasis on overall survival was secondary, contributing to the limited nature of the effect estimate of months-per-patient lived. No risk of bias was assessed per study during the compilation of the studies. Future directions for this research should include a more in-depth analysis using risk of bias tools, homogeneous analysis, and sensitivity assessment. This would help to examine whether these treatments allow some patients to live longer or more patients to live more months. A more detailed effect estimate could provide a confident representation of the true effect of neoadjuvant treatment.
At the time of our review 10 years ago, gemcitabine was the primary first-line neoadjuvant chemotherapy. Data around FOLFIRINOX originated in 2011 when it was used to treat high performance patients with metastatic pancreatic cancer, resulting in a superior 11-month median survival when compared to gemcitabine monotherapy. 53 In a recent review, preoperative FOLFIRINOX was found to be associated with 33-month median survival following resection and 75% R0 resection rate, positioning itself as an attractive and promising avenue for neoadjuvant therapy. 54 In one study, disease free survival was found to be comparable between FOLFIRINOX (19 months) and GEMOX therapy (23 months). 29 When comparing by percent survived, after 1 year following surgery, patients treated with FOLFIRINOX were alive 6% more of the time compared to Gemcitabine monotherapy and 24% more than CRT monotherapy. 19 Despite the superior median survival and negative margin rate compared to gemcitabine, concerns have been raised regarding the neurotoxicity, increased adverse events, progression in poor responsiveness and interruption or suspension of therapy.15,51 Concerns of progression and generalized toxicity remain, considering this review found that more patients make it to an operation and undergo resection in upfront resection groups compared to neoadjuvant therapy by wide margins.
Neoadjuvant immunotherapy is a unique option for pancreatic adenocarcinoma. Due to the pancreas’s tumor microenvironment, pancreatic cancer can be particularly protective against immune system activation due to the naturally immunosuppressive nature of the pancreas. This could potentially prove difficult for targeted immune therapy, however overcoming this suppression barrier is a primary goal. 55 A study investigated potential adjuvant vaccination of patients with pancreatic cancer using atezolizumab with FOLFIRINOX to trigger neoantigen immunogenicity in T-cells. Half of the 16 patients treated with the vaccine had a successful neoantigen T-cell response, which was not correlated to margin status. 19% of patients did not finish the vaccination regimen due to disease progression. 56 This is a novel finding which opens the path for new and more intensive immunotherapies. Additionally, the ever-changing scope of defining resectability may offer an expansion of what is considered resectable or not given the ever-evolving field of minimally invasive Surgery for advanced pancreatic disorders, yet no studies exist in examining the resectability of locally advanced pancreatic cancer when utilizing the robotic platform.
The argument has been made prior that tumor responsiveness to neoadjuvant therapy is an indicator of initial resectability, stating that candidates who responded poorly to neoadjuvant therapy and had progression from resectable or borderline would not have been good candidates for resection regardless.57,58 However, the data regarding this is thin, as responsiveness to neoadjuvant therapy as a marker for resectability neglects the importance of curative resection. Resection remains the only cure for pancreatic adenocarcinoma, and placing patients through an oncological stress test with marginally efficacious treatment is a mistreatment to an entire generation of patients. 59 The utilization of neoadjuvant therapy should be approached carefully and be considered for unresectable, locally advanced and metastatic diseases, given the evidence of its association with improved performance status, decreased tumor markers, and addressal of lymph node metastases as opposed to primary tumors which show little radiological evidence for downstaging.50,60
Conclusion
This study has examined medical literature on neoadjuvant therapy and its effects on oncological outcomes in patients with borderline resectable pancreatic cancer. We affirm 10 years later that the data on neoadjuvant therapy has improved yet remains marginally efficacious and comparable to the upfront resection approach when examining curative intent analysis. Patients undergoing neoadjuvant therapy make it to an operation in detrimentally decreased numbers, have fewer resections, and comparable negative margins to patients receiving an upfront resection. Research conducted on toxicity, responsiveness, and patient-reported factors between different neoadjuvant regimens in intention-to-treat cohorts is necessary to address the possible diffraction of outcomes in the literature. We encourage practitioners to reassess the usage of neoadjuvant therapy as a first-line approach for patients who are resectable and potentially resectable with borderline resectable pancreatic cancer. We expect that the introduction of modified and immunotherapy regimens or minimally invasive procedures may offer new opportunities to patients and providers to expand the first-line resectability of patients with advanced pancreatic disease.
Footnotes
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Sharona Ross is a consultant for Intuitive Surgical (Sunnyvale, CA) and Ethicon (Cincinnati, OH). Dr. Sharona Ross receives educational grants for her Women in Surgery Career Symposium from Intuitive Surgical and Medtronic (Minneapolis, MN). The remaining authors have no conflicts of interest or financial ties to disclose.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
