Journal Logo


A Call for Caution in Overinterpreting Exceptional Outcomes After Radical Surgery for Pancreatic Cancer

Let the Data Speak

Datta, Jashodeep MD; Wilson, Gregory C. MD; D’Angelica, Michael I. MD; Katz, Matthew H. G. MD§; Maithel, Shishir K. MD; Merchant, Nipun B. MD; Ahmad, Syed A. MD

Author Information
doi: 10.1097/SLA.0000000000004471
  • Free

“Tumor biology is king; selection is queen; and technical maneuvers are the princes and princesses trying to usurp the throne, sometimes with temporary apparent victories, and usually to no long-term avail.” Perhaps no other disease lends itself to the fundamental tenets of surgical oncology—eloquently outlined by Blake Cady over 20 years ago—as pancreatic ductal adenocarcinoma (PDAC).1 Fortunately, multimodality therapies for the treatment of localized and/or metastatic PDAC have shown steady, albeit small, incremental improvements since publication of this seminal article.2–8 Paramount to these developments has been increasingly effective modern combination chemotherapy regimens delivered in the metastatic, adjuvant, and now neoadjuvant settings; safety and technical mastery of pancreatectomy; and growing appreciation of the molecular underpinnings in PDAC. Despite these advances, the recent celebration of 5-year survival in PDAC eclipsing the double-digit threshold serves as a sobering reminder of its overwhelmingly lethal and unrelenting biology.

In parallel with these developments, there has been a noticeable surge in the literature of case series championing radical operations for borderline resectable/locally advanced pancreatic cancer (BR/LAPC) and even metastatic disease.9–12 These series from high-volume institutions describe the “feasibility” of arterial resections (AR), combined arterial and venous resections, and metastasectomy for PDAC in selected patients, and their apparent associations with improved survival. Although the reported outcomes following these extreme measures highlight commendable technical prowess, they serve to further reinforce the underestimated contributions of physiologic and biologic selection to their results. These contributions may be underappreciated by the broader surgical community and warrant closer dissection. The goal of the present article is to not only serve as a call to exercise caution in overinterpreting these data, but also to provide a constructive framework to critically evaluate such studies.

Although series describing the renaissance of radical operations for PDAC tout their safety and feasibility, they often overlook the challenge of contextualizing these outcomes relative to a meaningful population-at-risk—an epidemiologic principle known as the “denominator problem.”13 By reporting only on treated/resected patients, these studies neglect the inherent selection biases which dramatically alter the denominator from which percentages are calculated, resulting in an augmented proportion of patients with favorable outcomes.14,15 For instance, in a study reporting AR during pancreatectomy, median overall survival (OS) of patients undergoing upfront AR is dramatically worse compared with patients undergoing resection after neoadjuvant therapy (16.6 vs 53.6 months, P = 0.038).9 While acknowledging that most surgeons in modern practice would not perform AR in BR/LAPC patients without administering neoadjuvant therapy (NAT), these data illustrate the points of selection and denominators nicely. Importantly, in these reports, the reader is not provided the true denominator of patients who initiated NAT for BR/LAPC but never came to surgical resection, or even the number of patients with arterial involvement who were never referred for surgery. Without this information, it is impossible to interpret the relative effectiveness of AR after NAT. The reported outcomes discount those who developed metastatic disease, experienced decline in performance status, and suffered chemotherapy-related attrition while on NAT, thereby falsely enriching the denominator for patients able to be technically resected—an inherently favorable cohort both biologically and physiologically. Furthermore—with median follow-up of only 19 months—the upfront surgery and NAT curves seem superimposed at early timepoints, separating only after several patients in the NAT cohort are censored due to inadequate follow-up. Therefore, interpretation of the 54-month median OS in the NAT cohort must be tempered. It is entirely possible that the 2 survival curves would overlap, as they did in the first 12 months, with longer follow-up (Fig. 1). While acknowledging the technical feasibility of such radical operations, it is imperative to know the true denominators of study populations, better understand the selection biases inherent in referral for such operations, perform intention-to-treat survival analyses, and insist on adequate longitudinal follow-up data to understand the true impact of these radical approaches.

Overall survival analysis comparing patients undergoing arterial resection who underwent surgery first compared with those receiving neoadjuvant therapy (NAT). This is an example to illustrate the concept that knowledge of true denominators of patients initiating NAT in an intent-to-treat fashion and longitudinal follow-up information are required to interpret comparative effectiveness between these 2 cohorts. Yellow circles indicate censored data points at which the survival curves, otherwise superimposed, seem to separate.

Another critical factor when interpreting such studies is accounting for biologic selection: do more extensive operations for advanced tumors with seemingly favorable biology alter their natural history? As systemic therapies improve, we are observing responses after NAT that seemed improbable a few years ago. Despite such responses, it remains unclear if radical operations offer oncologic benefits that not only outweigh their heightened morbidity/risk profile and often diminished postoperative quality of life, but whether they also outperform the oncologic outcomes seen after continued systemic and/or less invasive locoregional therapies. To illustrate these concepts, we highlight a series on 128 metastasectomies for oligometastatic PDAC between 2001 and 2014—an average of 9 patients/yr and a small denominator compared to the cumulative population-at-risk over 14 years. One-third of patients underwent resection of distant aortocaval lymph nodes and two-thirds underwent hepatectomy with an identical median OS of 12.3 months. In particular, in the cohort selected for hepatectomy, 30-day mortality and actual 5-year OS were near-identical—4.3% and 5%, respectively.12 In another single-institution series reporting on 118 ARs for LAPC over 27 years (∼4.3 cases/yr), perioperative mortality was 5%, median OS 13 months, and few 5-year survivors.16 Juxtapose these outcomes with the median OS of 26 months and similarly few 5-year survivors reported in LAPC patients with radiographically stable but unresected disease after FOLFIRINOX and consolidation chemoradiotherapy.17 While we acknowledge that cross-study comparisons are imperfect, it seems that radical surgery may not improve the disease trajectory in patients with seemingly indolent biology and, at worst, might actually harm patients. Moreover, despite these technically spectacular “local” victories, the overwhelming majority of patients ultimately succumb to distant disease. As these examples illustrate, there is substantial overlap between the Venn diagrams of patients with locally advanced/metastatic PDAC who may be technically resectable with radical resections and similar patients with responsive/stable disease following systemic therapies in whom the disease course would remain unaffected without radical resection.

In the era of improved responses with modern systemic regimens, it is entirely possible that there is a subset of patients (eg, stable/responsive disease after NAT for extended periods without distant progression) that may benefit from radical surgery having passed the biologic “test of time.” However, our current selection tools are unable to accurately identify such patients. Therefore, as a surgical community, we must weigh the absolute benefits of radical surgery with the price our patients pay to achieve them. A recent study revealed some sobering perioperative statistics: of 111 ARs, 1 in 8 patients died within 90 days of surgery, over half suffered grade 3 or higher complications, and 1 in 6 underwent reoperation.9 Data from another series examining BR/LAPC patients receiving NAT showed that more extensive vascular reconstructions were associated with worse median survival (no AR: 24.2 months, AR: 14.8 months, vascular bypass: 9.8 months). Given that there was no way to biologically or physiologically separate these tumors/patients a priori, it is possible that the morbidity and quality of life impairments—rarely reported in such studies—associated with radical surgery negatively altered the natural history of disease in these patients. As such, we must accept that tumor geography is less important than tumor biology, and a closer dissection of the outcomes discussed herein underscores the competing tensions between these 2 often disparate concepts.

The repeated publication of data championing radical operations for PDAC, therefore, unintentionally creates heuristic frameworks that reinforce our collective surgical bias—that a chance to cut is a chance to cure. The reality is that outcomes reported from experienced tertiary referral centers will likely not be replicated in other institutions. The related concern is that others may use such data as a green light to adopt them into practice settings less equipped to provide the multidisciplinary resources required for these patients or lacking the ability to rescue patients from adverse complications synonymous with radical surgery, resulting in even worse outcomes. Furthermore, this genre of data is tantalizing because it convinces us that we as a surgical community are making progress in the field—the ultimate confirmation bias. The reality is that the demonstrated technical prowess of increasingly radical resections has done little to move the needle forward in this disease with respect to survival outcomes.18 Instead, our collective efforts must focus on improving understanding of its biological heterogeneity and therapeutic vulnerabilities. We need to not only develop more effective and less toxic systemic therapies but also discover novel biomarkers of therapeutic response/resistance. As such, frequent integration of transcriptomic signatures (eg, classical and basal/squamous19) predicting therapeutic responses in PDAC,20 or identification of molecular predictors of localized non-metastatic disease failure (eg, intact SMAD421) may enable strategic application of radical surgery to subsets of patients—no matter how small—who might benefit.

We must take care, however, to not allow these biological and statistical considerations to undercut surgical innovation. There has been undoubted progress in minimal access surgery and making radical surgery safer. Notwithstanding, while exceptional technical feats should be recognized, the reality is that only a select few patients are candidates for such operations and come to resection when selected for such a path. Even when radical surgery in select patients is temporarily successful, it is highly debatable whether the ensuing outcomes surpass that which could have been achieved with less morbid systemic therapies. Meaningful advances in the care of PDAC patients ultimately rests on developing effective biomarkers to distinguish those who will benefit from extensive operations, and altering the trajectory of its biology at the cellular level and not just in the gross resection of visible disease, no matter how elegant or enticing it may be.


1. Cady B. Basic principles in surgical oncology. Arch Surg 1997; 132:338–346.
2. Neoptolemos JP, Dunn JA, Stocken DD, et al. Adjuvant chemoradiotherapy and chemotherapy in resectable pancreatic cancer: a randomised controlled trial. Lancet 2001; 358:1576–1585.
3. Oettle H, Post S, Neuhaus P, et al. Adjuvant chemotherapy with gemcitabine vs observation in patients undergoing curative-intent resection of pancreatic cancer: a randomized controlled trial. Jama 2007; 297:267–277.
4. Neoptolemos JP, Moore MJ, Cox TF, et al. Effect of adjuvant chemotherapy with fluorouracil plus folinic acid or gemcitabine vs observation on survival in patients with resected periampullary adenocarcinoma: the ESPAC-3 periampullary cancer randomized trial. JAMA 2012; 308:147–156.
5. Neoptolemos JP, Palmer DH, Ghaneh P, et al. Comparison of adjuvant gemcitabine and capecitabine with gemcitabine monotherapy in patients with resected pancreatic cancer (ESPAC-4): a multicentre, open-label, randomised, phase 3 trial. Lancet 2017; 389:1011–1024.
6. Conroy T, Hammel P, Hebbar M, et al. FOLFIRINOX or gemcitabine as adjuvant therapy for pancreatic cancer. N Engl J Med 2018; 379:2395–2406.
7. Conroy T, Desseigne F, Ychou M, et al. FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer. N Engl J Med 2011; 364:1817–1825.
8. Von Hoff DD, Ervin T, Arena FP, et al. Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine. N Engl J Med 2013; 369:1691–1703.
9. Tee MC, Krajewski AC, Groeschl RT, et al. Indications and perioperative outcomes for pancreatectomy with arterial resection. J Am Coll Surg 2018; 227:255–269.
10. Loveday BPT, Zilbert N, Serrano PE, et al. Neoadjuvant therapy and major arterial resection for potentially reconstructable arterial involvement by stage 3 adenocarcinoma of the pancreas. HPB (Oxford) 2019; 21:643–652.
11. Klaiber U, Schnaidt ES, Hinz U, et al. Prognostic factors of survival after neoadjuvant treatment and resection for initially unresectable pancreatic cancer. Ann Surg 2019; doi: 10.1097/SLA.0000000000003270. Online ahead of print.
12. Hackert T, Niesen W, Hinz U, et al. Radical surgery of oligometastatic pancreatic cancer. Eur J Surg Oncol 2017; 43:358–363.
13. Chunara R, Wisk LE, Weitzman ER. Denominator issues for personally generated data in population health monitoring. Am J Prev Med 2017; 52:549–553.
14. D’Angelica M. Hepatic resection for metastatic breast cancer: an exercise in selection bias. HPB (Oxford) 2016; 18:631–632.
15. Di Maio M, Audisio M, Cardone C, et al. The use of not-negative conclusions to describe results of formally negative trials presented at oncology meetings. JAMA Oncol 2020.
16. Bachellier P, Addeo P, Faitot F, et al. Pancreatectomy with arterial resection for pancreatic adenocarcinoma: how can it be done safely and with which outcomes?: a single institution's experience with 118 patients. Ann Surg 2020; 271:932–940.
17. Sadot E, Doussot A, O’Reilly EM, et al. FOLFIRINOX induction therapy for stage 3 pancreatic adenocarcinoma. Ann Surg Oncol 2015; 22:3512–3521.
18. Winter JM, Brennan MF, Tang LH, et al. Survival after resection of pancreatic adenocarcinoma: results from a single institution over three decades. Ann Surg Oncol 2012; 19:169–175.
19. Collisson EA, Sadanandam A, Olson P, et al. Subtypes of pancreatic ductal adenocarcinoma and their differing responses to therapy. Nat Med 2011; 17:500–503.
20. Aung KL, Fischer SE, Denroche RE, et al. Genomics-driven precision medicine for advanced pancreatic cancer: early results from the COMPASS trial. Clin Cancer Res 2018; 24:1344–1354.
21. Iacobuzio-Donahue CA, Fu B, Yachida S, et al. DPC4 gene status of the primary carcinoma correlates with patterns of failure in patients with pancreatic cancer. J Clin Oncol 2009; 27:1806–1813.

arterial resection; biology; biomarkers; borderline resectable; chemotherapy; denominator; intention to treat; locally advanced; metastasectomy; pancreatic cancer; radical surgery

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.