Clinical trialObservationalNot yet recruiting
Preoperative CA 19-9 and the Pattern and Timing of Recurrence After Pancreatoduodenectomy (Whipple Operation) for Pancreatic Cancer
Correlation of Preoperative Serum CA 19-9 With the Pattern and Timing of Recurrence After Pancreatoduodenectomy for Pancreatic Ductal Adenocarcinoma: A Retrospective Cohort Study
- Outcome
- Ongoing
- Phase
- Observational
- Status
- Not yet recruiting
- Study type
- Observational
- Start
- 1 December 2026
- Expected end
- 30 April 2027
- Enrollment
- 180 participants
- Sites
- 2
- Primary outcomes
- Anatomical pattern of first recurrence, by index preoperative CA 19-9 group (>37 vs ≤37 U/mL); Time to first recurrence, with death without recurrence as a competing event, by index preoperative CA 19-9 group
- Sponsor type
- Academic / other
- Results posted
- No
- Last update
- 18 November 2026
These fields come from the trial’s registry entry. A summary of its population, design, endpoints and results, each tied to a quote, has not been written for this trial yet.
Ask this document
Answers come only from this document’s own text, as highlighted quotes. If it doesn’t say, you are told so instead of given a guess.
Try
Source text
Brief summary public domain (ClinicalTrials.gov)
Surgery to remove the head of the pancreas (pancreatoduodenectomy, also called the Whipple operation), usually followed by chemotherapy, is the only treatment that offers long-term survival for cancer in the head of the pancreas (pancreatic ductal adenocarcinoma). Even so, the cancer comes back (recurs) in most patients. Where it comes back first (around the operation site, in the liver, in the lining of the abdomen, in the lungs or elsewhere) and how soon it comes back both affect survival, the choice of further treatment and how patients should be followed up. CA 19-9 is a blood test measured routinely before this operation. A high level is known to be linked with shorter survival, but whether it also predicts where and when the cancer will recur is uncertain and, to the investigator's knowledge, has not been studied in patients treated in Egypt. Jaundice, which is common in these patients, can raise CA 19-9 falsely, so this study uses only CA 19-9 values measured when the blood bilirubin level was close to normal (2 mg/dL or less). This is a retrospective study of existing medical records. It includes adults who had a Whipple operation for this cancer at the Hepatopancreatobiliary and Surgical Oncology Unit, Liver and GIT Hospital, Minia University, Egypt, between January 2017 and December 2024, with follow-up information from Minia Oncology Centre. No patient will receive any test or treatment for the study, and no care will be changed. Patients will be compared according to their preoperative CA 19-9 level (37 U/mL or less versus more than 37 U/mL). The two main questions are whether the preoperative CA 19-9 level is linked to (1) the site of the first recurrence and (2) the time from surgery to recurrence. A recurrence will count only when confirmed by scans or biopsy, never by a rising CA 19-9 alone, and it will be classified by independent doctors who do not know the patients' CA 19-9 results. Other questions include recurrence within 6 and 12 months of surgery, survival, and whether CA 19-9 adds useful information to the routine findings available before surgery. The investigator expects that patients with a high CA 19-9 will more often recur first in the liver or the lining of the abdomen and will recur sooner, whereas patients with a normal level will recur later and more often only near the operation site. The findings may help tailor follow-up after surgery, identify patients who might benefit from chemotherapy before surgery, and identify patients whose cancer is likely to come back only at the operation site, who might be considered for further local treatment.
Detailed description public domain (ClinicalTrials.gov)
BACKGROUND AND RATIONALE Pancreatoduodenectomy followed by adjuvant systemic therapy is the only treatment sequence associated with long-term survival for pancreatic ductal adenocarcinoma (PDAC) of the pancreatic head, yet most patients recur even after a margin-negative resection and completed adjuvant chemotherapy. The site of first failure and its timing carry separate prognostic information. Hepatic and peritoneal failure carry a worse outlook than isolated local recurrence, which may occasionally be amenable to re-resection or stereotactic radiotherapy, and 12 months of recurrence-free survival separates early from late recurrence with markedly different survival after recurrence. Nodal status alters the timing of recurrence without materially altering its anatomical distribution, so this study treats pattern and timing as two separate primary questions. Serum carbohydrate antigen 19-9 (CA 19-9; sialyl Lewis a) is the only serum biomarker in routine clinical use for this disease, and an elevated preoperative value is consistently associated with shorter overall and recurrence-free survival. Three properties constrain its interpretation: approximately 5-10% of people are Lewis-negative non-secretors who cannot synthesise the antigen; biliary obstruction raises CA 19-9 independently of tumour burden; and no threshold above the conventional upper limit of normal (37 U/mL) has been universally adopted. Three gaps motivate the study. First, most evidence linking preoperative CA 19-9 to outcome uses composite endpoints (overall or recurrence-free survival) that collapse anatomically and biologically distinct failure events; whether the marker predicts where a patient will fail has been examined in few series and, to the investigator's knowledge, never in an Egyptian or North African population. Second, studies of recurrence pattern have generally treated death without recurrence as non-informative censoring, which misstates absolute risk; a competing-risk formulation is preferable. Third, confounding by obstructive jaundice and biliary drainage is rarely handled by an explicit, pre-specified rule for measuring the exposure. Egyptian practice also differs from the settings that dominate this literature: upfront surgery predominates for resectable disease, obstructive jaundice at presentation is common, preoperative biliary drainage is frequently required, and access to multi-agent adjuvant chemotherapy is variable. Because serum tumour markers are insufficient as a stand-alone criterion for recurrence, and because CA 19-9 is also the exposure, admitting a rising marker as evidence of recurrence would make any finding circular; recurrence therefore requires radiological or histological confirmation throughout. OBJECTIVES
Detailed description public domain (ClinicalTrials.gov)
* Primary: to determine whether the index preoperative serum CA 19-9 concentration is associated with (1) the anatomical pattern of first recurrence after pancreatoduodenectomy for PDAC and (2) the time from operation to first recurrence, with death without recurrence treated as a competing event. * Secondary: association with early (≤12 months) and very early (≤6 months) recurrence; site-specific cumulative incidence of first recurrence at 6, 12, 24 and 36 months, overall and by CA 19-9 stratum; association with recurrence-free, overall and post-recurrence survival; the prognostic contribution of CA 19-9 relative to established pathological predictors and its incremental discrimination over routinely available preoperative variables; and description of perioperative outcomes, adjuvant treatment delivery and surveillance completeness. * Exploratory: a data-driven CA 19-9 threshold separating early from late recurrence (not to be presented as a validated cut-off); characteristics and outcomes of patients with CA 19-9 below 2 U/mL (presumed Lewis-negative non-secretors); and whether the CA 19-9 to total bilirubin ratio retains prognostic information in patients for whom no bilirubin-qualified value was obtainable. HYPOTHESES Among patients who recur, an elevated index CA 19-9 (\>37 U/mL) is associated with a higher relative frequency of distant-only first recurrence, particularly hepatic and peritoneal, and a lower relative frequency of local/locoregional-only first recurrence. An elevated index CA 19-9 is associated with a higher subdistribution hazard of recurrence, with an anticipated hazard ratio of approximately 1.8 relative to a normal value. Null hypotheses: the distribution of first-recurrence patterns and the subdistribution hazard of recurrence do not differ across CA 19-9 strata. Hypotheses are directional, but all tests are two-sided at alpha 0.05. DESIGN, SETTING AND PERIOD Single-institution, two-site, retrospective observational cohort study with a competing-risk outcome framework. Operative, anaesthetic, laboratory, radiology and synoptic pathology records come from the Hepatopancreatobiliary and Surgical Oncology Unit and the Department of Pathology, Liver and GIT Hospital, Faculty of Medicine, Minia University; adjuvant treatment, surveillance imaging, tumour-marker trends, recurrence documentation and vital status come from Minia Oncology Centre. Operations performed from 1 January 2017 (establishment of the dedicated hepatopancreatobiliary service, from which synoptic pathology and structured operative records are consistently available) to 31 December 2024 are eligible. Follow-up is censored on 31 August 2026, giving every patient at least 20 months of potential observation. Cumulative incidence is reported to 36 months, and beyond 36 months only where at least 15 patients remain at risk. No aspect of clinical care is altered, and no investigation, sample or patient contact is undertaken for the study other than a single vital-status enquiry where approved. The study follows the STROBE statement for cohort studies. The statistical analysis plan (protocol Section 10) is fixed before database lock; any later change will be dated, documented and reported as a post-hoc amendment. STANDARD OF CARE DURING THE ACCRUAL WINDOW (CONTEXT ONLY; NOT ALTERED BY THE STUDY)
Detailed description public domain (ClinicalTrials.gov)
Pancreatic-protocol contrast-enhanced CT, with MRI/MRCP where the CT is equivocal and endoscopic ultrasound with tissue sampling where a diagnosis is required before treatment; CA 19-9 and CEA at diagnosis and, in jaundiced patients, after biliary decompression; preoperative biliary drainage (endoscopic stenting by default) for cholangitis, severe or symptomatic jaundice, anticipated delay to surgery or nutritional optimisation; review at the joint hepatopancreatobiliary and oncology meeting; pancreatoduodenectomy with standard ISGPS lymphadenectomy, with mesopancreatic excision and an artery-first approach where indicated and venous resection where tumour contact precludes a clear plane; axial slicing with multicolour margin inking and synoptic reporting against the AJCC eighth edition dataset, applying the 1 mm rule; referral for adjuvant chemotherapy within 8 to 12 weeks (gemcitabine monotherapy, gemcitabine with capecitabine, or modified FOLFIRINOX in selected fit patients); and surveillance by clinical review, CA 19-9 and contrast-enhanced CT of the chest, abdomen and pelvis every 3 months for 2 years, then every 6 months to 5 years, with additional imaging for symptoms or a rising marker. EXPOSURE: INDEX PREOPERATIVE CA 19-9 The index value is the last serum CA 19-9 measured within the 30 days preceding operation at a contemporaneous total bilirubin ≤2 mg/dL (≤34 µmol/L), contemporaneous meaning drawn within 72 hours of the CA 19-9 sample. In patients who underwent preoperative biliary drainage it is the post-drainage value satisfying that criterion; pre-drainage values are recorded separately and never used as the exposure. Where several values qualify, the one closest to operation is used; where none qualifies, the patient enters the secondary cohort and is analysed only in sensitivity analysis S2. Assays are electrochemiluminescence immunoassays; the platform, reference interval and reporting limits in force in each calendar year are tabulated, and any change of platform is entered as a covariate in a sensitivity analysis. Values beyond the limits of quantification are assigned that limit, their number is reported, and a multiple-imputation sensitivity analysis is pre-specified. The exposure is analysed in four pre-specified forms: * Ordinal: non-secretor \<2; normal 2-37; moderately elevated \>37-200; markedly elevated \>200 U/mL. Used for descriptive stratification, cumulative incidence functions, Gray's test and pattern tables, with a test for trend across the three producing levels. * Binary: \>37 versus ≤37 U/mL, with non-secretors in the ≤37 stratum (sensitivity analysis S1 excludes them). Primary statistical contrast for both primary outcomes. * Continuous: log2(CA 19-9 + 1), giving the change in hazard per doubling; non-linearity assessed by a restricted cubic spline with knots at the 10th, 50th and 90th centiles. * Data-driven threshold (exploratory only): maximally selected rank statistics with the Hothorn-Lausen exact correction, labelled as requiring external validation. OUTCOME DEFINITIONS Pattern of first recurrence (mutually exclusive categories):
Detailed description public domain (ClinicalTrials.gov)
* Local/locoregional only: new or progressively enlarging soft tissue in the pancreatic remnant or any anastomosis (pancreaticojejunostomy, hepaticojejunostomy, gastro- or duodenojejunostomy), the pancreatic bed and uncinate groove, perivascular soft tissue of the superior mesenteric artery or coeliac axis, the portal vein/superior mesenteric vein groove, or regional lymph node stations 8, 12, 13, 14 or 17. * Liver only: one or more new hepatic lesions with imaging characteristics of metastasis, or proven hepatic metastasis, without disease elsewhere. * Peritoneal only: peritoneal or omental nodularity, omental caking, or new ascites with positive cytology or unequivocal accompanying peritoneal nodules, without disease elsewhere; new ascites alone does not qualify. * Lung or other distant only: new pulmonary nodules consistent with metastasis, or metastasis at any other distant site (bone, adrenal, distant soft tissue, or non-regional nodes including para-aortic station 16), without disease elsewhere. Station 16 recurrence is deliberately classified as distant nodal, consistent with the AJCC eighth edition. * Multiple synchronous sites: disease in two or more categories on the same study or on studies within 30 days of the first evidence of recurrence. Time to recurrence: months from operation to the first imaging study or biopsy demonstrating recurrence; where recurrence is established by comparison across two studies, the date of the later, confirmatory study. Competing event: death from any cause without documented recurrence. Censoring: alive and recurrence-free at last documented contact or at the follow-up cut-off, whichever is earlier. Early, very early and late recurrence: at or before 12 months, at or before 6 months, and after 12 months, respectively. RECURRENCE ASCERTAINMENT AND BLINDED ADJUDICATION Every recurrence requires radiological or histological confirmation. Acceptable evidence is: biopsy or cytology showing adenocarcinoma consistent with the resected primary (the procedure date is the recurrence date); an unequivocal new distant lesion on CECT or MRI, new or progressive locoregional soft tissue ≥10 mm, or a new regional nodal mass ≥10 mm in short axis; a previously indeterminate lesion that has unequivocally enlarged on a study performed at least 6 weeks later (the confirmatory study date is the recurrence date); or focal FDG avidity on PET/CT at a compatible site with a structural correlate on CT or MRI. Not acceptable alone: a rising serum CA 19-9, new or worsening symptoms, clinical suspicion, new ascites without corroboration, or postoperative inflammatory change at the operative bed on early imaging. Patients with biochemical progression but no confirmed recurrence are censored at their last recurrence-free imaging study and counted in a pre-specified sensitivity analysis. Two clinicians independent of the study (a consultant radiologist and a consultant hepatopancreatobiliary surgeon) independently review all surveillance imaging reports, and the images where equivocal, in packets from which every CA 19-9 value has been removed, recording the fact, date and pattern of recurrence. Disagreement on the fact, on the date by more than 30 days, or on the pattern is resolved by a third independent clinician (a consultant medical oncologist) whose decision is final. Agreement on pattern is reported by Cohen's kappa together with the number of cases requiring adjudication. Whether recurrence was first detected on symptom-triggered or scheduled imaging is recorded. COVARIATES
Detailed description public domain (ClinicalTrials.gov)
Collected in six domains, with permitted values and source documents defined in the case report form and data dictionary: demographic and baseline (age, sex, body mass index, ASA class, ECOG performance status, Charlson age-comorbidity index, new-onset versus longstanding diabetes, smoking); presentation (jaundice, peak bilirubin, biliary drainage modality and interval, cholangitis, albumin, C-reactive protein, haemoglobin, neutrophil and platelet counts with derived ratios, CEA); imaging and staging (CT tumour diameter, NCCN resectability category, radiological nodal status); operative (approach, procedure type, extent of lymphadenectomy, venous or arterial resection, pancreatic texture and duct diameter, operative time, blood loss, transfusion, coded surgeon); pathology (tumour size, grade, pT and pN, nodes examined and involved with lymph node ratio, lymphovascular and perineural invasion, margin status with each margin recorded individually, extrapancreatic extension, background PanIN or IPMN); and postoperative and follow-up (complications graded by Clavien-Dindo, the comprehensive complication index and ISGPS/ISGLS definitions, length of stay, adjuvant therapy, surveillance imaging dates and findings, surveillance CA 19-9 values used descriptively only, recurrence details and vital status). Staging follows the AJCC eighth edition, margin assessment the Leeds Pathology Protocol, and lesion measurement RECIST version 1.1. DATA SOURCES, EXTRACTION AND QUALITY CONTROL Data are drawn from the operative logbook and theatre register, hospital electronic and paper records, the clinical pathology laboratory information system (including assay platform and reference intervals), the radiology information system and picture archive, the Department of Pathology archive (blocks and slides available for re-review where a synoptic field is missing or internally inconsistent) and the Minia Oncology Centre record. The principal investigator extracts data onto a purpose-built electronic case report form with range checks, permitted-value lists and mandatory fields, in three passes (surgical, pathological and oncological), after a pilot on 15 records. A random 20% of records is extracted independently by a clinician blinded to the investigator's extraction (Cohen's kappa for categorical and intraclass correlation coefficient for continuous variables; a kappa below 0.80 for any key variable triggers re-extraction of that variable across the whole cohort). Every exposure value and every recurrence date undergoes 100% source verification by an independent second reader. Automated logic checks run before lock (operation before recurrence, recurrence before death, a bilirubin for every index CA 19-9, involved nodes not exceeding nodes examined, and no recurrence without a source-document reference); every query is logged and closed, and the dataset is locked before any inferential analysis, with the lock documented by date, file checksum and signature. SAMPLE SIZE
Detailed description public domain (ClinicalTrials.gov)
The cohort is a fixed population: every eligible patient operated within the accrual window is included. Using Schoenfeld's formula with a hazard ratio of 1.8 for CA 19-9 \>37 versus ≤37 U/mL, 35% of patients with a bilirubin-qualified value ≤37 U/mL, 75% cumulative recurrence, two-sided alpha 0.05 and 80% power, 100 recurrence events are required (133 patients; 153 allowing 15% attrition). A minimum of 155 analysable patients (approximately 116 events) is the threshold below which inferential analysis will not proceed; this gives 86% power for a hazard ratio of 1.8, and the smallest hazard ratio detectable with 80% power is 1.73. Institutional volume is expected to yield 170 to 190 eligible patients. The study is not powered for modest differences within individual pattern categories or for prediction-model development: power for early recurrence (odds ratio 2.30) is approximately 70%, multivariable models are capped at ten events per variable, penalised estimation is used where a pattern has fewer than 30 events, and no category with fewer than 10 events is modelled. Such findings are reported as estimates with confidence intervals and described as hypothesis-generating. STATISTICAL ANALYSIS Analysis populations: the principal population (all eligibility criteria met, bilirubin-qualified index CA 19-9, adjudicated recurrence status) for all primary analyses; the pattern population (adjudicated first recurrence) for the multinomial analysis; the secondary population (no bilirubin-qualified value obtainable) for sensitivity analysis S2 only; and the early-recurrence set (at least 12 months of follow-up or an event within 12 months). Descriptive analyses: characteristics by CA 19-9 stratum with standardised mean differences rather than p-values; the distribution of the index value on raw and log2 scales with the proportion at each reporting limit; perioperative outcomes, pathology, adjuvant delivery and surveillance completeness by stratum; and follow-up duration by the reverse Kaplan-Meier method. Primary analysis 1 (pattern): pattern-specific cumulative incidence by the Aalen-Johansen estimator with competing events (recurrence in another category; death without recurrence) at 6, 12, 24 and 36 months, compared across strata by Gray's test; Fine-Gray subdistribution hazard and cause-specific Cox models for each pattern with at least 10 events (Firth penalisation below 30 events), reported together with their interpretive difference; and multinomial logistic regression of pattern among patients who recur (reference: local/locoregional only), adjusted for the confounder set. Categories with fewer than 10 patients are collapsed in a fixed order (lung-only into other-distant-only; other-distant-only into peritoneal-only if still sparse); local/locoregional-only, liver-only and multiple-site categories are never collapsed, and any collapsing is reported. Primary analysis 2 (timing): cumulative incidence of recurrence of any pattern with death without recurrence as the competing event, by stratum (Gray's test); primary effect estimate the subdistribution hazard ratio for \>37 versus ≤37 U/mL from a multivariable Fine-Gray model adjusted for the confounder set, with a cause-specific Cox model fitted in parallel; the hazard ratio per doubling of CA 19-9 with a restricted cubic spline of adjusted hazard ratio against concentration; and restricted mean time to recurrence at 24 months by stratum.
Detailed description public domain (ClinicalTrials.gov)
Secondary analyses: multivariable logistic regression for early and very early recurrence in the early-recurrence set; Kaplan-Meier estimation with the log-rank test and multivariable Cox regression for recurrence-free and overall survival; descriptive post-recurrence survival by pattern and stratum; likelihood-ratio tests between nested Cox models and change in Harrell's C-index for CA 19-9 versus the pathological predictors; and the difference in time-dependent area under the curve at 12 and 24 months, with bootstrap confidence intervals, between a preoperative model (age, sex, Charlson age-comorbidity index, CT tumour size, radiological nodal status, resectability category) with and without CA 19-9. Covariate selection is causal, not statistical. The minimal sufficient adjustment set is age, sex, Charlson age-comorbidity index, obstructive jaundice with preoperative biliary drainage, and year of surgery as a period effect. Pathological variables (pT and tumour size, pN and lymph node ratio, grade, lymphovascular and perineural invasion, margin status) are treated as mediators and are not adjusted for in the primary model; they enter a pre-specified secondary model, the comparison estimating how far the association operates through conventional pathology. Adjuvant chemotherapy receipt and regimen are modelled as a time-varying covariate to avoid immortal time bias, with a 90-day landmark analysis as a check. Postoperative complications and treatment received at recurrence are descriptive only. Stepwise selection and selection on univariable p-values are not used. Model diagnostics and missing data: proportional hazards assessed by scaled Schoenfeld residuals, with a time-varying coefficient where violated and restricted mean survival time differences at 12 and 24 months if the exposure itself violates the assumption; influential observations by dfbeta, with re-analysis without any observation that changes the primary estimate by more than 15%; variance inflation factor threshold of 5; functional form by martingale residuals. The exposure and primary outcomes are complete in the principal population by construction. For other variables: complete-case analysis below 5% missingness (checked against the imputed result); multiple imputation by chained equations with 20 imputations between 5% and 30% (imputation model including all analysis variables, the Nelson-Aalen cumulative hazard and the event indicator; estimates combined by Rubin's rules); and no imputation above 30%, the variable being reported descriptively only. The missing-at-random assumption is discussed for each imputed variable, with a pattern-mixture sensitivity analysis where missingness is likely to depend on unobserved values. Pre-specified sensitivity analyses (all reported, whether or not they agree with the primary analysis): * S1: presumed Lewis-negative non-secretors (\<2 U/mL) excluded from the reference stratum. * S2: re-analysis of the secondary population using the CA 19-9 to bilirubin ratio, with bilirubin as a covariate. * S3: restriction to patients with at least 80% protocol-adherent surveillance imaging. * S4: interval-censored analysis, recurrence occurring between the last negative and the first positive study. * S5: exclusion of recurrences detected by symptom-triggered rather than protocol-driven imaging. * S6: 90-day landmark analysis for adjuvant chemotherapy in place of the time-varying covariate. * S7: multiple imputation for assay values reported beyond the limits of quantification. * S8: primary contrast repeated at published alternative thresholds of 100, 200 and 500 U/mL. * S9: operating surgeon added as a shared frailty term. * S10: exclusion of patients operated in 2017 and 2018 (learning-curve and period effects).
Detailed description public domain (ClinicalTrials.gov)
Subgroups, multiplicity and model performance: five pre-specified exploratory subgroups (margin status R0 versus R1; nodal status N0 versus N1/N2; receipt of adjuvant chemotherapy; obstructive jaundice at presentation; grade, well or moderately versus poorly differentiated) are examined by interaction terms and shown in a forest plot. No alpha adjustment is applied between the two primary objectives, which address distinct outcomes; the Benjamini-Hochberg procedure controls the false discovery rate at 5% within the family of pattern-specific hazard models, with unadjusted and adjusted p-values reported; secondary, subgroup, sensitivity and exploratory analyses are not alpha-adjusted and are labelled hypothesis-generating. Discrimination is quantified by Harrell's C-index and time-dependent area under the curve at 12 and 24 months, calibration by predicted versus observed cumulative incidence in deciles of predicted risk, and optimism by bootstrap resampling with 1000 replicates repeating the whole modelling procedure; performance is described as apparent, internally validated performance in a single-centre cohort, with no claim of external validity. Analyses use R version 4.4 or later (survival, cmprsk, riskRegression, prodlim, mstate, rms, mice, maxstat, nnet, timeROC, coxphf, ggplot2) with version-controlled code. BIAS AND LIMITATIONS Pre-specified mitigation: selection bias from excluding patients without a bilirubin-qualified value (excluded patients characterised; S2); detection bias from more intensive surveillance of patients with an elevated marker (surveillance completeness reported by stratum; S3 to S5); incorporation bias (imaging or histological confirmation of every event; adjudicators blinded to all marker values); information bias (structured case report form, 20% duplicate extraction, 100% source verification of exposure and primary outcomes, standardised external definitions); immortal time bias (time-varying adjuvant chemotherapy; S6); competing-risk misestimation (Aalen-Johansen and Fine-Gray methods, Kaplan-Meier reserved for composite endpoints); confounding by indication (biliary drainage in the adjustment set); assay drift (platform recorded per year; year of surgery adjusted for); attrition (structured tracing; loss reported by stratum); and overfitting (events-per-variable cap, penalisation, bootstrap optimism correction). Limitations: retrospective single-centre design (association and temporal sequence, not causation; unmeasured confounding such as molecular subtype and circulating tumour DNA status); interval-scheduled imaging, so times to recurrence are upper bounds; lower sensitivity of CT for peritoneal recurrence; non-secretor status inferred rather than genotyped; heterogeneous adjuvant regimens; and limited power for pattern-specific inference. ETHICS, CONSENT AND DATA PROTECTION
Detailed description public domain (ClinicalTrials.gov)
Approved by the Medical Research Ethics Committee / Institutional Review Board, Faculty of Medicine, Minia University (MUFMIRB approval number 2216:10/26, 3 October 2026). The study is conducted in accordance with the Declaration of Helsinki (2013 revision) and Egyptian Clinical Medical Research Law No. 214 of 2020 and its Executive Regulation No. 927 of 2022; no investigational product or device is involved. Individual informed consent is waived because the research involves no more than minimal risk, the waiver does not adversely affect participants' rights or welfare, the research could not practicably be carried out otherwise (a substantial proportion of the cohort is deceased, and requiring consent would introduce severe non-random selection bias among those with the outcome of interest), and a study summary is available to the department and to any surviving patient or family member who enquires. Where a surviving patient or next of kin is telephoned solely to establish vital status, verbal permission is sought at the start of the call and nothing is requested beyond vital status and the fact and date of any recurrence. Each patient receives a sequential study identifier; the analysis dataset contains no name, hospital number, national identification number, address or telephone number. The linkage file is encrypted, accessible only to the principal investigator and destroyed 5 years after publication; data are held on encrypted institutional storage, never on personal devices, personal cloud accounts or removable media; access is logged; and records are retained for 10 years from publication. No individual-level data leave Minia University, no specimens are transferred, and no data are shared with any commercial entity. Access to Minia Oncology Centre records is subject to that centre's institutional permission. ORGANISATION, FUNDING AND DISSEMINATION Single-investigator study: the principal investigator holds every study role and is the guarantor of the data and analysis. Four functions are delegated to clinicians independent of the study (two blinded outcome assessors, a blinded adjudicator, and an independent second reader for duplicate extraction and source verification), who take no part in design, analysis, interpretation or reporting. The study is unfunded and uses existing institutional records; the investigator has no financial or non-financial interest in the CA 19-9 assay, any assay manufacturer or any therapy referenced. Results will be submitted for peer-reviewed publication irrespective of their direction, reported according to STROBE, and added to this record within 12 months of database lock. Substantive amendments (eligibility, exposure or primary outcome definitions, primary analysis) will be approved by the ethics committee before implementation and recorded here.
Papers 9
- BackgroundPreoperative predictors for early and very early disease recurrence in patients undergoing resection of pancreatic duct… DOI 10.1016/j.hpb.2021.09.004↗
- BackgroundPredicting Adverse Pathologic Features and Clinical Outcomes of Resectable Pancreas Cancer With Preoperative CA 19-9. DOI 10.3389/fonc.2021.651119↗
- BackgroundTiming But Not Patterns of Recurrence Is Different Between Node-negative and Node-positive Resected Pancreatic Cancer. DOI 10.1097/sla.0000000000003123↗
- BackgroundDefining and Predicting Early Recurrence in 957 Patients With Resected Pancreatic Ductal Adenocarcinoma. DOI 10.1097/sla.0000000000002734↗
- BackgroundImplications of the Pattern of Disease Recurrence on Survival Following Pancreatectomy for Pancreatic Ductal Adenocarci… DOI 10.1245/s10434-018-6558-7↗
- BackgroundPatterns, Timing, and Predictors of Recurrence Following Pancreatectomy for Pancreatic Ductal Adenocarcinoma. DOI 10.1097/sla.0000000000002234↗
- BackgroundSystematic review on the role of serum tumor markers in the detection of recurrent pancreatic cancer. DOI 10.1016/j.hpb.2017.11.009↗
- BackgroundCarbohydrate Antigen 19-9 Elevation in Anatomically Resectable, Early Stage Pancreatic Cancer Is Independently Associat… DOI 10.1016/j.jamcollsurg.2016.02.009↗
- BackgroundSerum CA 19-9 as a Biomarker for Pancreatic Cancer-A Comprehensive Review. DOI 10.1007/s13193-011-0042-1↗