Analysis of diagnostic pathway and therapeutic delay for patients with pancreatic adenocarcinoma: a French retrospective study

Abstract

Background:

Pancreatic adenocarcinoma (PA) is a growing public health concern and is expected to become the second leading cause of cancer-related death in Europe by 2030. Despite therapeutic advances, prognosis remains poor, with stage at diagnosis the main determinant of outcome, highlighting the need for earlier diagnosis.

Objectives:

The primary objective was to identify factors associated with the delay between first symptoms and treatment initiation. Secondary objectives included the impact of this delay on tumor stage, progression-free survival (PFS), and overall survival (OS).

Design:

This study is a single-center retrospective observational cohort.

Methods:

We retrospectively reviewed medical records of patients with newly diagnosed PA between January 2018 and December 2020 to analyze diagnostic and treatment pathways, delays between first patient-reported symptoms and treatment initiation, and their association with tumor stage, PFS, and OS.

Results:

A total of 275 consecutive patients were included. The mean time from the first symptom-to-treatment initiation was 120.1 days, significantly shorter in cases of jaundice (82.6 days) and longer for diabetes (179.6 days; p < 0.001). In multivariate analysis, greater weight loss (β = 2.44, 95% confidence interval: 0.75–4.14; p = 0.005) was linked to longer delays. No correlation was found between time from the first symptom-to-treatment initiation and tumor stage (p = 0.13), OS (p = 0.69) in the overall population, and PFS (p = 0.77) for patients with first-line palliative chemotherapy.

Conclusion:

The delay between the first symptom and PA treatment initiation remains long, but no survival impact was observed with shorter delays. Symptomatic disease is diagnosed and treated faster, but with no significant impact on survival. These findings underline the need to evaluate new strategies to accelerate PA diagnosis.

Plain language summary

How delays from first symptoms to treatment start impact survival of patients with pancreatic adenocarcinoma in France

Pancreatic adenocarcinoma is associated with a poor prognosis and will become the second leading cause of cancer-related death in Europe by 2030. Early diagnosis is important, but the time from first symptoms to the start of treatment has been poorly evaluated. In this study, we analyzed medical records of 275 patients diagnosed with pancreatic adenocarcinoma between 2018 and 2020 to understand treatment delays and what factors influenced these delays. Median delay from first symptom to the start of treatment was about 4 months and patients with jaundice were diagnosed and treated faster (around 3 months), while those whose first symptom was diabetes/poor general health waited longer for treatment to start (around 6 months). Importantly, shorter delays from first symptoms to treatment did not result in better survival. Nevertheless, these results need confirmation by larger multicenter series since our series is single-center, retrospective and with a limited number of patients. The relatively long time from symptom report to treatment start highlights the need to find new strategies to diagnose pancreatic cancer quickly and start specific treatment earlier.

Keywords

healthcare pathway overall survival pancreatic adenocarcinoma treatment delay

Introduction

Pancreatic adenocarcinoma (PA) accounts for 90% of pancreatic tumors.¹ In 2022, it ranked as the 12th most common cancer in terms of incidence and the 6th in terms of mortality, with 510,992 new cases and 467,409 deaths worldwide.² Incidence is steadily increasing and PA will become the second leading cause of cancer-related mortality in Europe by 2030.^3,4 The main prognostic factor is resectability (i.e., R0 resection), which is possible in only 10%–20% of cases, due to the aggressiveness of the disease and nonspecific initial symptoms, potentially leading to diagnostic delays and diagnosis at advanced unresectable stage.¹

The most common signs at PA diagnosis include weight loss, asthenia, anorexia, abdominal pain, with or without radiation to the back, gastrointestinal disorders, and jaundice. Due to these nonspecific symptoms, it is often the general practitioner (GP) who sees these patients first.⁵ In cases with symptoms suggesting pancreatic disease, it is recommended to perform an abdominopelvic computed tomography scan (CT-scan).¹ If the imaging suggests PA, the patient is quickly referred to a specialized physician for tumor staging, according to TNM and National Comprehensive Cancer Network classification, before any treatment is decided.⁶ The staging assessment, in addition to abdominopelvic CT-scan, includes a chest CT-scan and liver magnetic resonance imaging to identify metastatic lesions, and an endoscopic ultrasonography to stage non-metastatic PA.⁷ Furthermore, endoscopic biliary drainage should be considered in cases with biliary obstruction, and a tumor biopsy is mandatory in cases with metastatic/localized non-resectable tumors before palliative or induction treatment can be implemented.⁸ Thus, the healthcare pathway of PA patients requires coordination among many physicians, including GPs, gastroenterologists, oncologists, surgeons, and radiologists, to obtain a diagnosis and define the appropriate treatment during multidisciplinary tumor board meetings.

Even though it has not been clearly demonstrated, reducing the time to PA diagnosis and treatment could increase rates of curative surgery and decrease the proportion of patients with best supportive care alone, thereby improving overall survival (OS). Several studies have analyzed the delay between first symptoms and treatment start.^5,9–11 The REPERE study retrospectively explored the healthcare pathway of patients with metastatic PA in France. The median delay between the first symptoms and the pathological diagnosis ranged from 41 to 65 days, and this delay tended to be shorter in cases with jaundice and longer in cases with abdominal pain.⁵ In the prospective English SYMPTOM study, the average delay between symptom onset and diagnosis was 117 days, with a shorter delay in cases with jaundice or anorexia (85 and 114 days, respectively), while the delay increased when the first symptoms were diabetes, anxiety, or depression.¹¹ A retrospective single-center study suggested that although the diagnosis delay is significantly shorter for patients with jaundice, there was no significant impact in terms of resectability rates and OS.¹⁰ By contrast, a multicenter retrospective study (n = 324) showed that a delay of more than 25 days between the last medical consultation and the radiological examination was associated with a significant decrease in 1-year OS, but the impact of other delays (from diagnosis to treatment initiation) on survival was not evaluated.⁹

This study aims to describe the diagnostic pathway and time to care for patients with PA, to evaluate factors associated with time to treatment, and to evaluate the impact of delays on patients’ survival.

Methods

The reporting of this study complies with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cohort studies.¹²

Study population

We conducted a retrospective study at Poitiers University Hospital between January 15, 2018, and December 20, 2020. The date of inclusion was defined as the date of pancreatic cancer diagnosis. For patients without histological confirmation, the inclusion date corresponded to the multidisciplinary tumor board meeting at which the diagnosis of PA was established by a panel of experts and systemic treatment was decided. To ensure the inclusion of all consecutive patients and minimize selection bias, two complementary data sources were used: the multidisciplinary tumor board meetings database and the electronic chemotherapy prescription software (ChimioWeb^©, Computer Engineering, Paris, France). In France, all cancer patients must be registered at multidisciplinary tumor board meetings even if no treatment was given (best supportive care alone). Data were collected retrospectively from the date of the first consultation at Poitiers University Hospital until the date of death, or until the end of follow-up for alive patients (i.e., January 12, 2023). Dates of first symptoms were based on patient self-report and documented in the medical records. Dates of consultations with GPs or specialists were subsequently identified through referral letters and available medical correspondence.

The main inclusion criteria were all consecutive patients with PA, whether histologically proven (biopsy or surgical specimen) or not (for these patients, PA diagnosis was validated by experts during multidisciplinary tumor board meetings), regardless of tumor stage, and aged 18 years or more. Exclusion criteria were patients mostly managed outside Poitiers University Hospital and non-adenocarcinoma tumors (pancreatic neuroendocrine tumors and other pancreatic lesions, including pancreatic metastases). No formal a priori sample size or power calculation was performed, as this was an exploratory retrospective study that included all consecutive patients with newly diagnosed PA at our center between January 2018 and December 2020.

Patients’ and tumor characteristics

Patients’ (age, sex, weight, and Eastern Cooperative Oncology Group-Performance Status (ECOG-PS)) and tumor characteristics (resection of the primary tumor and TNM stage) were collected. Biological data including tumor markers (carcinoembryonic antigen (CEA) and carbohydrate antigen (CA 19-9)) and neutrophil to lymphocyte ratio (NLR)) were also analyzed. All dates and places of patient care were carefully collected using all medical reports (first symptom, first consultation, first imaging, pathological diagnosis, and first treatment).

Treatment and outcomes were collected at each line of chemotherapy, including the chemotherapy regimen, the start and end dates for each line of treatment, and tumor progression. Other treatments, like radiotherapy, chemoradiotherapy, and/or surgery, were also recorded.

This study was approved by our Committee for Ethics and Scientific Review of Health Data Research (CESRDS, opinion no. 2025-07-04).

Statistical analyses

All characteristics were described for the overall population. Quantitative variables were described using medians, and standard deviations (SD), and qualitative variables were described with frequencies and percentages.

The diagnostic and therapeutic delays for each patient were measured using the time from the first symptom to the date of the first consultation with any physician (T1), the time from the first consultation to the radiological suspicion of PA (T2), the time from the radiological suspicion of PA to the pathological diagnosis (T3), and the time from the pathological diagnosis to the start of treatment (T4; Figure 1). The start date of treatment was defined as the date of pancreatic surgery, or the start date of chemotherapy or radiotherapy, whichever occurred first. For patients with no cancer treatment, the date of the decision to provide best supportive care alone was chosen as the start date for treatment.

Figure 1.

Definitions of time intervals.

The primary objective was to assess the association between the delay from first symptoms to treatment initiation (T1 + T2 + T3 + T4, i.e., symptom-to-treatment delay) and OS in the overall population. OS was defined as the time from pathological diagnosis to death from any cause. Alive patients at the end of the follow-up were censored. Progression-free survival (PFS) was defined as the time from treatment initiation to disease progression or death, whichever occurred first. The PFS analysis was limited to patients receiving first-line palliative chemotherapy. The cut-off date for survival analysis was January 12, 2023. Median follow-up was estimated using the reverse Kaplan–Meier method. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test.

Associations between delays and survival outcomes were assessed using univariate Cox proportional hazards models. Factors associated with the symptom-to-treatment delay were determined using univariate analyses (Mann–Whitney test for qualitative variables and Spearman correlation for quantitative variables). Variables with a p-value <0.05 in univariate analysis were selected for inclusion in the multivariate model using multivariate linear regression (ordinary least squares) to identify factors independently associated with treatment delay, considered as a continuous outcome. No automated variable selection procedure (e.g., stepwise selection) was applied. Results are reported as β coefficients along with their 95% confidence intervals (CIs). β coefficients represent the mean difference in treatment delay associated with each variable. A p-value of less than 0.05 was considered statistically significant, and all tests were two-sided. All analyses were done using Statview^© software (SAS Institute, Cary, NC, USA).

Results

Patients’ and tumor characteristics at diagnosis

Altogether, 543 patients were registered in the multidisciplinary tumor board meetings database for PA, and 509 were registered in ChimioWeb^© software for PA between January 1, 2018, and December 31, 2020, at Poitiers University Hospital. Duplicates were removed, resulting in a total of 498 patients. Among these, 223 patients were excluded mostly because they had a non-adenocarcinoma tumor, the date of diagnosis was outside the inclusion period, and/or most of the treatment was provided in another hospital. Finally, 275 patients with PA were included (Figure 2).

Figure 2.

Flow chart.

Median age at PA diagnosis was 71.0 years, 51.3% were women, and most patients had an ECOG-PS of 0 or 1 (76.8%). At diagnosis, median albumin was 38.0 g/L, median bilirubin was 13.0 µmol/L, median CA 19-9 was 599.0 UI/L, and median NLR was 3.7. Primary tumors were mostly located in the pancreatic head (49.8%). Tumors were classified as resectable (13.1%), borderline (7.3%), locally advanced (20.0%), and metastatic (59.6%), according to the National Comprehensive Cancer Network criteria (Table 1 and Figure 2).⁶

Table 1.

Patients’ and tumor characteristics at diagnosis.

Characteristics at diagnosis	All patients (n = 275)
Median age (years, (min–max))	71.0 (40.0–90.0)
Sex (n, %)
Female	141 (51.3%)
Male	134 (48.7%)
Place of initial care^a (n, %)
Poitiers University Hospital	125 (45.5%)
Peripheral hospital	79 (28.7%)
Private clinic	58 (21.1%)
Other region	13 (4.7 %)
Body mass index (kg/m²) (median, IQR)	23.7 (21.3–27.2)
Weight loss (% of usual self-reported weight, median (IQR)	8.5% (2.6–13.9)
ECOG-PS (n, %)
0	59 (21.5%)
1	152 (55.3%)
2	45 (16.4%)
3	18 (6.5%)
4	1 (0.4%)
Primary tumor location (n, %)
Head	135 (49.8%)
Body	77 (28.4%)
Both body and tail	20 (7.4%)
Tail	39 (14.4%)
Missing	4
First medical consultation (n, %)
General practitioner	181 (68.0%)
Emergency unit	54 (20.3%)
Specialized physician^b	31 (11.7%)
Missing	9
Blood tests at diagnosis (median (IQR))
NLR	3.7 (2.6–6.7)
CEA (µg/L)	5.7 (3.0–16.0)
CA19.9 (U/L)	599.0 (82.0–3117.0)
Albumin (g/L)	38.0 (33.5–42.5)
Bilirubin (µmol/L)	13.0 (8.0–140.3)
LDH (U/L)	347.5 (301.3–412.3)
First symptom of PA (n, %)
Pain	103 (37.5%)
Poor general health^c	86 (31.3%)
Jaundice	28 (10.2%)
Diabetes mellitus	19 (6.9%)
Digestive symptoms	11 (4.0%)
Cholestatic syndrome	6 (2.2%)
Other	9 (3.3%)
Fortuitous	13 (4.7%)

It refers to the location of the first consultation for suspected PA since all patients were then treated at Poitiers University Hospital.

Gastroenterologists, medical oncologists, radiation therapists, or digestive surgeons.

PGH includes asthenia, involuntary weight loss and/or loss of appetite.

CA19.9, cancer antigen 19-9; CEA, carcinoembryonic antigen; ECOG-PS, Eastern Cooperative Oncology Group-Performance Status; IQR, interquartile range; LDH, lactate dehydrogenase; NLR, neutrophil to lymphocyte ratio; PA, pancreatic adenocarcinoma; PGH, poor general health; SD, standard deviation.

The most frequent first symptoms were pain (37.5%), followed by poor general health (PGH; 31.3%), and jaundice (10.2%). At the beginning of the care pathway, 68.0% of patients had their first consultation with their GP, 20.3% went to a hospital emergency unit, and 11.7% had a consultation with a specialized physician (Table 1).

Therapeutic management

Among the 275 patients, 228 (82.9%) had treatment for PA that included chemotherapy, radiotherapy, and/or surgery. Among the 36 patients with a resectable tumor, 31 had pancreatic surgery (86.1%), 27 had an R0 resection, and 24 had adjuvant chemotherapy. Twenty-four patients had a tumor recurrence (77.4%) after surgery, and 22 of these were able to receive first-line palliative chemotherapy (91.7%; Supplemental Figure 1). Among the 20 patients with a borderline tumor, 1 had upfront surgery, 19 had induction chemotherapy, and 14 of these had secondary pancreatic resections (Supplemental Figure 2). Thirteen patients had a post-operative recurrence (86.7%), and 9 of these were able to receive first-line palliative chemotherapy (69.2%).

For the 55 patients with a locally advanced tumor, 7 had best supportive care alone, and 48 patients (87.3%) received induction chemotherapy (Supplemental Figure 3). After induction chemotherapy, 2 patients had surgery, 13 had palliative chemoradiotherapy, 15 had palliative chemotherapy alone, and 18 continued their induction chemotherapy as the tumor response was insufficient to allow surgery. These 18 ultimately received best supportive care alone with no further treatment. Most of the induction chemotherapy was 5FU-based regimens (45.8%: Folfirinox, Folfiri, Folfox, or LV5FU2). Among the 164 patients with metastatic PA, 127 received first-line chemotherapy (77.4%), while 37 patients (22.6%) had best supportive care alone (Supplemental Figure 4).

Treatment delay and factors associated with treatment delay

The mean delay from initial symptom-to-treatment (Time 1 + 2 + 3 + 4) was 120.1 ± 90.0 days, with a minimal delay of 9.0 days and a maximum delay of 843.0 days (Figure 3). The mean time between the initial symptom and the first appointment with a physician was 34.3 ± 45.4 days (Time 1). After this consultation, radiological imaging was performed in a mean time of 30.1 ± 54.8 days (Time 2), and a pathological diagnosis was provided in 19.4 ± 20.1 days (Time 3). After the pathological diagnosis, a mean delay of 36.3 ± 21.9 days was needed to start the treatment (Time 4).

Figure 3.

Time intervals for diagnosis and treatment, in the overall population and according to the first symptom. The first symptom “Cholestasis” is not shown as it represents a small subgroup (n = 6), likely not informative in this context.

The mean delay from the first symptom-to-treatment varied depending on the type of the first symptom (p < 0.001). Patients with jaundice as the first symptom were diagnosed and treated in a significantly shorter time (82.6 ± 61.6 days) than were patients with pain (119.6 ± 95.6 days), PGH (131.7 ± 81.7 days), and diabetes (179.6 ± 118.2 days). This long delay was related to a prolonged Time 1 of 55.2 ± 47.3 days for PGH and a prolonged Time 2 of 101.8 ± 109.2 days for diabetes (Figure 3). Among patients with jaundice, the shortest delay was observed between symptom onset and the first medical consultation (8.5 ± 5.6 days), and between radiological suspicion of PA and pathological diagnosis (12.6 ± 18.1 days).

In univariate analysis, predictive factors for the symptom-to-treatment delay were the type of initial symptoms (p < 0.0001), the type of physician consulted initially (p = 0.01), tumor location (p = 0.02), weight loss (p < 0.0001), and ECOG-PS (p = 0.008), as well as lymphocyte count (p = 0.004), NLR (p = 0.0006), bilirubin (p = 0.003), and albumin (p = 0.0005) levels (Table 2). There was no correlation between tumor stage and symptom-to-treatment time (p = 0.14). In multivariable analysis, only substantial weight loss remained significantly associated with a longer symptom-to-treatment delay (β = 2.44, 95% CI: 0.75–4.14; p = 0.005).

Table 2.

Predictors of time between the first symptom and start of treatment, in univariate and multivariate analyses.

Characteristics	Univariates analysis		Multivariate analysis
Characteristics	Mean time (days)^a	p-Value	Regression coefficient, β (days)	(95% CI)	p-Value
Age	/	0.54	/		/
Sex			/		/
Female	135.7	0.95
Male	136.3
Center			/		/
Poitiers University Hospital	135.6	0.92
Peripheral hospital	131.9
Private clinic	141.4
Other region	140.3
First symptom^b
PGH (ref)	131.7	<0.0001	–	–	0.31^c
Pain	119.6		0.35	(−32.83, 33.54)	0.98
Jaundice	82.6		−7.69	(−56.16, 40.78)	0.76
Diabetes mellitus	179.6		24.18	(−29.63, 77.98)	0.38
Digestive symptoms	128.7		19.56	(−46.32, 85.43)	0.56
Cholestatic syndrome	115.0		−7.21	(−87.85, 73.43)	0.86
Other	65.6		−18.39	(−91.85, 55.07)	0.62
Fortuitous	102.5		−5.04	(−74.19, 64.11)	0.89
Loss of weight^b	/	<0.0001	2.44	(0.75, 4.14)	0.005
First consultation^b
Emergency unit (ref)	103.8	0.01	–	–	0.15^c
GP	139.0		21.52	(−10.36, 53.40)	0.18
Specialized physicians	135.9		17.84	(−28.20, 63.88)	0.45
ECOG-PS^b
4 (ref)	17	0.008	–	–	0.11^c
3	101.8		96.00	(−79.61, 271.60)	0.28
2	129.2		106.26	(−63.75, 276.26)	0.22
1	149.7		120.43	(−49.45, 290.30)	0.16
0	118.3		99.71	(−73.43, 272.86)	0.26
Tumor location^b
Head (ref)	127.0	0.02	–	–	0.10^c
Multiple	124.8		−28.98	(−76.88, 18.93)	0.23
Body	157.4		11.69	(−20.41, 43.79)	0.47
Tail	116.4		−25.90	(−63.84, 12.05)	0.18
Stage at diagnosis			/		/
Resectable	110.5	0.14
Borderline	127.2
Locally advanced	148.2
Metastatic	138.7
Tumor resection			/		/
Yes	125.1	0.30
No	138.3
Lymphocytes	/	0.004	/		/
NLR^a	/	0.0006	−5.25	(−5.25, 0.83)	0.15
Bilirubin^a	/	0.003	−0.08	(−0.20, 0.05)	0.22
Albumin^a	/	0.0005	1.51	(−0.98, 4.01)	0.23
CA 19.9	/	0.66	/		/

All variables are considered at the diagnosis. β coefficients represent the mean difference in treatment delay associated with each variable. Positive β coefficients indicate a longer delay, whereas negative β coefficients indicate a shorter delay, compared with the reference category.

Mean time (days) is reported for categorical variables only. For continuous variables, mean delay values are not shown, as their effects are evaluated as continuous predictors in the regression analysis.

Variable included in multivariate analysis.

The p-value shown for the reference category is the overall p-value for the variable (from the multivariable regression), while the p-values listed for each category correspond to the specific p-value for that category.

CI, confidence interval; NLR, neutrophil to lymphocyte ratio; PGH, poor general health.

Treatment delay and impact on survival

Median follow-up was 37.3 ± 1.9 months. At the end of follow-up, 219 deaths had occurred. Median OS from pathological diagnosis was 13.6 ± 0.78 months. The median OS according to the tumor stage was 8.3 ± 1.9 months for metastatic PA, 18.7 ± 0.9 months for locally advanced tumor, 32.2 ± 5.9 months for borderline tumor, and 33.8 ± 0.7 months for resectable PA (p < 0.0001). Overall, 207 patients had palliative first-line chemotherapy with a median PFS of 7.9 ± 0.2 months.

The symptom-to-treatment delay, analyzed as a continuous variable, had no impact on OS (p = 0.69), nor when analyzed as a binary variable (p = 0.36) in the overall population, and no impact on PFS (p = 0.77) for patients with palliative first-line chemotherapy.

Discussion

In this large retrospective series of PA patients, we identified factors associated with a longer delay between initial symptoms and the start of treatment. Factors included PGH and diabetes as the first symptoms, and in multivariate analysis, there was a greater weight loss. However, the symptom-to-treatment delay had no significant impact on OS or PFS.

The mean age at diagnosis in our cohort (71.0 years) is older than that in other series, due to the inclusion of all-comer, real-life PA patients in our study. For the same reason, the proportion of patients with ECOG-PS 0–1 in our study (76.8%) was lower than that in other published series.^13–16 Other initial characteristics of patients and tumors are consistent with the literature: tumors are mostly located in the pancreatic head, and mostly metastatic disease.^1,17

In our series, the mean delay between the first symptom and the initiation of treatment was 120.1 days. Isolated jaundice, the first symptom for 10.2% of patients and a well-recognized warning symptom of PA, led to significantly faster treatment initiation (82.6 days), as previously described. Indeed, the SYMPTOM study, a prospective English study, highlighted a significantly shorter time between symptom onset and pathological diagnosis in patients with jaundice, decreasing from 117 to 85 days, while the delay was longer when patients presented other first symptoms, such as diabetes or anxiety and depression disorders.¹¹ In the retrospective French study REPERE, Hammel et al. explored the healthcare pathway of patients with metastatic PA in France and diagnostic delays. The study involved 62 digestive oncologists or gastroenterologists, 300 GPs, and 200 patients. The median delay between the onset of the first symptoms and the final diagnosis was reduced by an average of 15 days in patients with jaundice (26–50 days) compared to 41–65 days in the overall population.⁵ This delay is consistent with our findings; we found 47.3 days for patients with jaundice and 83.8 days for the overall population (Time 1 + 2 + 3). It is interesting to note that, in our study, the time to the first consultation was shorter in patients with jaundice than in the overall population (8.5 vs 34.3 days), probably because jaundice is an easily identifiable symptom. The time saved was thus mostly between the onset of symptoms and the first consultation. Indeed, jaundice without fever/pain is a well-recognized warning sign of PA that requires rapid imaging and frequently biliary drainage. The presence of jaundice thus accelerates patients’ management and simultaneously allows pathological examinations (biliary stent by endoscopic retrograde cholangiopancreatography and biopsy by endoscopic ultrasonography). By contrast, in patients with diabetes or a deterioration in general health, a significantly longer therapeutic delay was observed, 179.6 and 131.8 days, respectively; a similar increase was observed in the SYMPTOM and REPERE studies.^5,6 Although it is recommended to investigate for PA in case of new-onset or decompensation of diabetes in at-risk patients, these patients experience a delay in diagnosis.¹⁸ In our study, pain was the most frequently reported initial symptom (37.5%), which is consistent with the findings of other studies (41%).⁵

It is worth noting that the factors that increased symptom-to-treatment delay included a better ECOG-PS, a low NLR, and substantial weight loss. Only one study has used multivariate analysis to look at factors associated with symptom-to-treatment delay, and the authors also identified loss of weight.¹² Substantial weight loss is considered a warning sign of cancer and should therefore accelerate the diagnosis of PA. In the same way, a low NLR is associated with less inflammatory and less symptomatic PA, which likely explains the longer symptom-to-treatment delay in this scenario, as also seen with better ECOG-PS status.¹⁹

Furthermore, our study revealed that the longer delay from the first symptom-to-treatment initiation did not significantly impact OS or PFS. This is counterintuitive, but given the poor prognosis of PA, whatever the stage, an association between early diagnosis and survival is difficult to demonstrate. Furthermore, the most aggressive diseases are generally more symptomatic and diagnosed earlier, but in fact have a worse prognosis. This result is consistent with the retrospective English study by Raptis et al.¹⁰ This study, conducted in a dedicated pancreatology unit between 1997 and 2002 and involving 355 patients, showed that delays to diagnosis or treatment did not impact OS. In multivariate analysis, the main predictive factor was the presence of a resectable tumor, which had a strong effect on OS (hazard ratio (HR) = 4.41, 95% CI: 2.4–8.1). Although the time from symptom onset to the referral letter was statistically associated with OS, its clinical impact was negligible (HR = 1.01, 95% CI: 1.001–1.002). These delays are not directly comparable to ours, given the specific care pathway in the English pancreatology unit, where access depended on a formal referral process. Moreover, it is possible that this lack of a significant difference was related to unidentified confounding factors, including patient selection at diagnosis, with the most severe cases having shorter treatment initiation delays. This association between longer delays to treatment initiation and survival is still a matter of debate, with contradictory studies published. Differences could be due to subjective, difficult-to-obtain information, especially the date of the first symptom, as well as retrospective data and different time periods evaluated.^20,21 In the study by Balzano et al., a delay of 25 days or more between the last medical consultation and the imaging-based suspicion of PA was associated with reduced patient OS (HR = 1.7, 95% CI: 1.2–2.3). However, it is difficult to compare these results with ours for several reasons; first, the time-points were different from those in our study, second, they analyzed all patients together with no distinction between metastatic and non-metastatic disease, which have quite different prognoses. Indeed, they did not collect data on the delay between the first symptom and the initial consultation with a physician (corresponding to our T1). Furthermore, the first time-point considered—the “last medical visit performed”—was not clearly defined.⁹

Our study revealed significantly varied symptom-to-treatment timelines, depending on the initial symptoms of the disease, thus underscoring the need to reduce delays between the first symptom and the first consultation with a physician and then imaging. To address this, a French non-interventional study was conducted to evaluate an accelerated care pathway called “One Day Diagnosis” (1DD).²² This care pathway consolidated essential steps, such as imaging, the anesthesia consultation, and the surgical evaluation, into a single day, thereby enabling an accurate diagnosis in 81.7% of patients. The program was implemented for patients with serious hepatobiliary or pancreatic diseases, including both cancer and non-cancerous diseases. The results showed that the time from patients’ referral to diagnosis was significantly shorter in the 1DD group than in patients following the conventional pathway (1 vs 15 days, p < 0.0004). However, the overall time from referral to treatment initiation did not significantly differ between the groups (69 days for the 1DD group vs 65 days, p = 0.69). The impact of this diagnostic interval on patients’ survival was not specifically assessed, but it may be limited, given that the total care timeline was not reduced.

A recent study by Kovacevic et al.²³ specifically examined the delay before surgery and potential tumor growth during this interval. They found that surgical treatment delay in upfront resectable patients did not appear to be associated with survival or the risk of recurrence; however, the optimal and maximal time to surgery and the ideal timing for follow-up imaging remain unclear. These results cannot be directly compared to ours, as their study included only patients undergoing surgery, whereas our cohort involved the full diagnostic and treatment pathway.

The limits of our study include the retrospective data collection, which may have led to information bias concerning the first symptoms. Consequently, less specific symptoms, such as digestive signs or depression, may have been underestimated during data collection. This study has several limitations, including its retrospective and single-center design, which may limit the external validity of the findings. As the cohort was derived from a single tertiary care institution, caution is warranted when generalizing the results to other settings. However, Poitiers University Hospital is a tertiary care academic medical center serving as a regional referral center for the Nouvelle-Aquitaine region. It manages patients referred from both primary and secondary care settings, with diagnostic and therapeutic approaches aligned with national guidelines, thereby supporting the comparability of its practices with those of other French university hospitals and cancer care centers, but it is less representative of patients in private centers. Another limitation of this study is the exclusion of a substantial proportion of the initial cohort. Of the 498 screened patients, 223 were excluded to ensure a homogeneous study population and reliable data collection. Exclusions mainly concerned diagnoses made outside the inclusion period and tumor subtypes with distinct biological behavior (neuroendocrine tumors, intraductal papillary mucinous neoplasm, pancreatic metastases originating from other primary tumors, and other non-ductal pancreatic tumors). These exclusion criteria were necessary to address the study objectives and did not introduce selection bias. Only excluded patients treated in multiple/other institutions may have introduced a selection bias, but these cases were limited in number and thus had a limited impact on the generalizability of the findings to the overall pancreatic cancer population. Although dates of first symptoms were systematically recorded, they were based on patient self-reporting and may have been subject to recall bias, which should be considered when interpreting time-to-diagnosis analyses. Another limitation of our study is the lack of a formal sample size or power calculation, inherent to its exploratory design. Nevertheless, all eligible patients were included during the study period, thus maximizing the representativeness of the sample. Although our study period included the early phase of the COVID-19 pandemic, most patients were diagnosed before its onset. Moreover, data from a contemporaneous multicenter French study found no significant impact of the pandemic on time to diagnosis or time to treatment initiation in PA, suggesting that the inclusion of this period is unlikely to have substantially biased our results. Finally, given the number of univariate analyses performed, the risk of type I error cannot be excluded, and results should be interpreted with caution. While OS and PFS are important clinical endpoints, their lack of significant association with the outcome in univariate analyses, combined with the fact that survival prognostic factors were beyond the primary scope of this study, led to their exclusion from the multivariable model.

Among the strengths of this study, it is worth noting the large sample size, with all consecutive patients with PA, which allowed us to create subgroups based on the initial symptom. Additionally, the combination of two databases enabled a comprehensive study of the overall population treated at Poitiers University Hospital with no missing patients. Finally, the analysis of therapeutic pathways according to tumor staging revealed that tumors were appropriately classified and treated since the treatment strategies appeared to comply with current guidelines, which strengthens the relevance of the results.

Conclusion

This study sheds light on the diagnostic pathways and on treatment timelines for patients diagnosed with PA, a topic with few published studies. Patients presenting with jaundice as their initial symptom appeared to experience shorter delays to diagnosis and treatment, suggesting the importance of recognizing other warning signs, like digestive disorders or asthenia. Nevertheless, while diagnostic and treatment timelines varied, our study found no significant association between time from symptom onset to treatment initiation and OS or PFS. This work highlights the complexity of PA diagnosis, which requires a multidisciplinary approach and early intervention strategies to meet the diagnostic challenges in the presence of nonspecific symptoms. The quest for improved strategies to enhance early detection and faster initiation of treatment remains paramount in the ongoing battle against PA. Unfortunately, there is still no non-invasive test for PA screening/diagnosis. Even though delays to the diagnosis and treatment of PA have a minor impact on OS, as demonstrated in this study, further research is needed to optimize the diagnostic and therapeutic timelines for PA.

Supplemental Material

sj-doc-1-tag-10.1177_17562848261428776 – Supplemental material for Analysis of diagnostic pathway and therapeutic delay for patients with pancreatic adenocarcinoma: a French retrospective study

Supplemental material, sj-doc-1-tag-10.1177_17562848261428776 for Analysis of diagnostic pathway and therapeutic delay for patients with pancreatic adenocarcinoma: a French retrospective study by Camille Evrard, Bérénice Vauchel, Claire Boyer, Aurélie Ferru, Raphael Olivier and David Tougeron in Therapeutic Advances in Gastroenterology

Supplemental Material

sj-docx-2-tag-10.1177_17562848261428776 – Supplemental material for Analysis of diagnostic pathway and therapeutic delay for patients with pancreatic adenocarcinoma: a French retrospective study

Supplemental material, sj-docx-2-tag-10.1177_17562848261428776 for Analysis of diagnostic pathway and therapeutic delay for patients with pancreatic adenocarcinoma: a French retrospective study by Camille Evrard, Bérénice Vauchel, Claire Boyer, Aurélie Ferru, Raphael Olivier and David Tougeron in Therapeutic Advances in Gastroenterology

Footnotes

Acknowledgements

We thank all physicians participating in this study. We would like to thank the patients and their families.

Declarations

ORCID iDs

Camille Evrard

David Tougeron

Supplemental material

Supplemental material for this article is available online.

References

Neuzillet

Gaujoux

Williet

, et al. Pancreatic cancer: French clinical practice guidelines for diagnosis, treatment and follow-up (SNFGE, FFCD, GERCOR, UNICANCER, SFCD, SFED, SFRO, ACHBT, AFC). Dig Liver Dis 2018; 50: 1257–1271.

Global Cancer Observatory: Cancer Today. International Agency for Research on Cancer, https://gco.iarc.who.int/today/

Ferlay

Partensky

Bray

More deaths from pancreatic cancer than breast cancer in the EU by 2017. Acta Oncol 2016; 55: 1158–1160.

Rahib

Smith

Aizenberg

, et al. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and pancreas cancers in the United States. Cancer Res 2014; 74: 2913–2921.

Hammel

Coriat

Lledo

, et al. Care pathway of patients with metastatic pancreatic cancer in daily practice in France: results from the REPERE national survey. Bull Cancer (Paris) 2017; 104: 321–331.

Tempero

MA.

NCCN guidelines updates: pancreatic cancer. J Natl Compr Cancer Netw 2019; 17: 603–605.

Marion-Audibert

A-M

Vullierme

M-P

Ronot

, et al. Routine MRI with DWI sequences to detect liver metastases in patients with potentially resectable pancreatic ductal carcinoma and normal liver CT: a prospective multicenter study. Am J Roentgenol 2018; 211: W217–W225.

Van der Gaag

Rauws

EAJ

van Eijck

CHJ

, et al. Preoperative biliary drainage for cancer of the head of the pancreas. N Engl J Med 2010; 362: 129–137.

Balzano

Laurent

Florence

A-M

, et al. Time interval from last visit to imaging diagnosis influences outcome in pancreatic adenocarcinoma: a regional population-based study on linked medico-administrative and clinical data. Ther Adv Med Oncol 2022; 14: 17588359221113264.

10.

Raptis

Fessas

Belasyse-Smith

, et al. Clinical presentation and waiting time targets do not affect prognosis in patients with pancreatic cancer. Surgeon 2010; 8: 239–246.

11.

Walter

Mills

Mendonça

, et al. Symptoms and patient factors associated with diagnostic intervals for pancreatic cancer (SYMPTOM pancreatic study): a prospective cohort study. Lancet Gastroenterol Hepatol 2016; 1: 298–306.

12.

Von Elm

Altman

Egger

, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370: 1453–1457.

13.

Wainberg

Melisi

Macarulla

, et al. NALIRIFOX versus nab-paclitaxel and gemcitabine in treatment-naive patients with metastatic pancreatic ductal adenocarcinoma (NAPOLI 3): a randomised, open-label, phase 3 trial. Lancet 2023; 402: 1272–1281.

14.

Conroy

Desseigne

Ychou

, et al. FOLFIRINOX versus gemcitabine for metastatic pancreatic cancer. N Engl J Med 2011; 364: 1817–1825.

15.

Von Hoff

Ervin

Arena

, et al. Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine. N Engl J Med 2013; 369(18): 1691–1703.

16.

De La Fouchardière

Malka

Cropet

, et al. Gemcitabine and paclitaxel versus gemcitabine alone after 5-fluorouracil, oxaliplatin, and irinotecan in metastatic pancreatic adenocarcinoma: a randomized phase III PRODIGE 65-UCGI 36-GEMPAX UNICANCER Study. J Clin Oncol 2024; 42: 1055–1066.

17.

Institut National du Cancer. Panorama des cancers en France—Edition 2022—Ref : PANOKFR2022, https://www.e-cancer.fr/Expertises-et-publications/Catalogue-des-publications/Panorama-des-cancers-en-France-Edition-2022 (2022, accessed 2 July 2023).

18.

Mizuno

Nakai

Ishigaki

, et al. Screening strategy of pancreatic cancer in patients with diabetes mellitus. Diagnostics (Basel, Switzerland) 2020; 10: 572.

19.

Cupp

Cariolou

Tzoulaki

, et al. Neutrophil to lymphocyte ratio and cancer prognosis: an umbrella review of systematic reviews and meta-analyses of observational studies. BMC Med 2020; 18: 360.

20.

Jooste

Dejardin

Bouvier

, et al. Pancreatic cancer: wait times from presentation to treatment and survival in a population-based study. Int J Cancer 2016; 139: 1073–1080.

21.

Lukács

Kovács

Csanádi

, et al. Benefits of timely care in pancreatic cancer: a systematic review to navigate through the contradictory evidence. Cancer Manag Res 2019; 11: 9849–9861.

22.

Cherkaoui

Seeliger

Faucher

, et al. Evaluation of an innovative care pathway in the diagnostic and therapeutic management of hepatobiliary and pancreatic pathologies: “one-day diagnosis.” J Pers Med 2022; 13: 12.

23.

Kovacevic

Ewertsen

Kristensen

, et al. Progression of resectable pancreatic ductal adenocarcinoma during surgical delay and effects on survival—a propensity score matched study. Pancreatology 2025; 25: 228–233.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB

0.21 MB