Multifocal UTUC Location and Prognosis: An Evidence and Registry-Data Workflow | QSevidence 2026
Research on multifocal upper tract urothelial carcinoma, or UTUC, requires location, laterality and timing as well as lesion counts.
Multifocal UTUC Location and Prognosis: An Evidence and Registry-Data Workflow | QSevidence 2026
For urologic oncology and registry teams, QSevidence can support an ai clinical decision support platform workflow connecting clinical questions, coding definitions and survival evidence for professional review.
Define the question and the status of the results
The proposed study concerns nonmetastatic multifocal UTUC, comparing renal-pelvis, ureteral and combined involvement across unilateral and bilateral patterns. Its aims are to describe location patterns, examine associations with overall survival (OS) and cancer-specific survival (CSS), and assess whether location improves an existing prediction model. Population registries can support association studies, but only if the intended exposure is measurable.
A reproducible SEER extraction session, patient-level analysis dataset, six-pattern algorithm, event tables and full model outputs have not been verified, nor has an original publication matching the stated statistics. Consequently, specific sample counts, hazard ratios, survival rates, interaction significance and model improvements are not presented as verified findings. This article develops the study and verification pathway; it does not establish that bilateral location independently causes worse survival.
Clinical multifocality, registered primaries and location are different constructs
Multifocality may describe several lesions on one side or involvement of both renal pelvis and ureter. Bilateral disease adds laterality and timing. One registered tumor is not a complete lesion map, and several primary records for one person do not automatically identify synchronous disease at presentation. Local extension, recurrence, metachronous disease and independent primaries require separate definitions.
The NCI urinary-site registration training emphasizes site-specific rules for determining single and multiple primaries. Historical SEER inquiry 20091110 illustrates how timing rules applicable to particular diagnosis years affect counting across urinary sites. It is an example for historical coding review, not a rule to apply indiscriminately to every period.
For a 2004-2016 cohort, investigators should examine the rules, fields and staging versions effective in each interval. Current rules should not simply be imposed retrospectively. Primary-site codes C65.9 and C66.9 plus laterality do not by themselves demonstrate complete ascertainment of six spatial patterns. Availability of lesion count, ureteral segment and concurrent involvement must be established from the actual dictionary and records.
Turn six proposed patterns into auditable exposures
| Pattern | Required information | Main ambiguity |
|---|---|---|
| Unilateral multifocal renal pelvis | Clinical or pathological evidence of distinct lesions within one renal pelvis | A single site code may not contain lesion counts |
| Unilateral multifocal ureter | Lesion count, segments and synchronous timing on one side | A ureter code is not a segment-level map |
| Unilateral renal pelvis and ureter | Concurrent involvement of both sites on the same side | Both sites may not generate separate primary records |
| Bilateral renal pelvis | Disease on both sides within a prespecified diagnostic window | Later disease is not baseline synchronous exposure |
| Bilateral ureters | Both ureters involved, with distinct-lesion information | Record counts are not lesion counts |
| Bilateral renal pelvis and ureter involvement | An explicit map of the sites affected on each side | Bilateral does not necessarily mean all four anatomical units |
Test the algorithm in an authorized data environment, followed by registration and uropathology review. Preserve classifiable, uncertain and unavailable states; missing information should not become unilateral or unifocal disease. If the SEER product cannot identify the intended patterns, narrow the question to measurable registered-site associations or use a multicenter cohort with imaging, endoscopic and pathological mapping. Do not manufacture six groups from insufficient fields.
Construct a patient-level cohort and fix time zero
The proposed population comprises patients diagnosed in 2004-2016 with malignant urothelial tumors originating in renal pelvis or ureter, no distant metastases and available survival information. Before extraction, specify the submission version, registry coverage, histological scope, behavior, metastasis definition and follow-up cutoff. Restriction to 8120/3 may omit other urothelial subtypes; a narrow code set should not be described as all UTUC.
The flow diagram should report tumor records, unique patients and exclusions at each step. Death-certificate-only or autopsy cases and unavailable survival or exposure information require explicit handling. Surgical eligibility must also be internally consistent. A radical-nephroureterectomy cohort needs a clear operation definition; a study allowing other procedures or no surgery should not inadvertently exclude those patients through a contradictory rule.
The synchronous window and baseline definition are particularly important. Classifying a person as bilateral from initial diagnosis using a contralateral tumor diagnosed later requires survival until that later event and can introduce guarantee-time bias. Depending on the question, use a baseline synchronous cohort, a prespecified landmark or a time-varying exposure. Explain the corresponding population and estimand rather than replacing temporal design with convenient patient-ID linkage.
Separate OS, CSS and competing-event questions
OS runs from the chosen origin to death from any cause, with survivors censored at last follow-up. CSS requires an explicit cancer-related death definition and handling of unknown causes. The SEER cause-specific death documentation describes classification using cause of death, tumor sequence and original site, with changes across versions. Attribution deserves particular review in people with multiple primaries; any cancer death should not automatically be assigned to the target UTUC.
Kaplan-Meier curves and log-rank tests can describe prespecified survival comparisons, while Cox models require proportional-hazards assessment. For the real-world absolute probability of UTUC death, other-cause death is a competing event; use cumulative-incidence estimation and competing-risk analyses aligned with the question. Cause-specific hazards, subdistribution hazards and cumulative death probabilities are not interchangeable quantities.
Review adjustment, missingness and predictive gain together
Candidate covariates include age, sex, race, diagnosis year, grade, T and N stage, surgery and recorded treatment. Stage changes require mappings or era-specific analysis, without casually mixing clinical and pathological stages. Bilateral cases also require a rule for selecting the grade and stage of a side or lesion. Taking the most advanced lesion may change the question and needs justification in the analysis plan.
Complete-case analysis does not inherently remove bias. Compare retained and excluded patients, describe missingness by group, and consider multiple imputation and sensitivity analyses under appropriate assumptions. Imputation cannot create lesion maps that were never collected. The SEER treatment-data limitations distinguish chemotherapy recorded as yes from no/unknown. The latter is not confirmed absence of treatment. Adjustment for this field does not adequately capture regimen, dose, indication or selection into treatment.
Prespecify location-by-stage and location-by-grade interactions, reporting event counts and intervals rather than searching small six-category subgroups for significance. Compare baseline and extended prediction models in the same sample with the same endpoint. Alongside the C-index, examine calibration, resampling-based optimism and independent validation. Improved discrimination would not establish that changing surgery or adding systemic treatment prolongs survival.
Interpret location within clinical risk assessment
The current EAU UTUC risk-stratification chapter notes variation in multifocality definitions and advises against using multifocality alone for risk stratification. A six-pattern study therefore belongs within stage, grade and broader clinical assessment, not a predetermined label of exceptionally poor prognosis for every bilateral case.
Field carcinogenesis, clonal spread and multiclonal origins remain mechanisms to investigate; a location-survival association cannot distinguish them. Bilateral disease raises renal-preservation, feasibility and patient-preference issues. Anatomical categories are not automatic instructions to extend resection. Surveillance and adjuvant-treatment decisions require guidelines, actual pathology and multidisciplinary assessment rather than unverified registry statistics.
Make the evidence workflow reproducible
QSevidence can help retrieve evidence around UTUC multifocality, location, synchronicity, registration rules and survival endpoints. Teams can compare definitions and data levels, synthesize conflicts and gaps, and trace recommendations back to guidelines and dictionaries. Registration, urology and statistical reviewers can then develop versioned variable definitions, a cohort flow, an analysis plan and a result-verification checklist. Evidence organization does not imply an already validated individual prognostic calculator.
A next-stage multicenter cohort could collect bilateral lesion maps, renal function, treatment timing and longitudinal follow-up, distinguishing synchronous and metachronous disease in advance. Molecular work would need specified sampling sites, clonality methods and independent validation. These steps are necessary to determine whether detailed location adds information beyond established stage and grade.
Frequently asked questions
How can QSevidence support preparation of a multifocal UTUC location study?
Teams can organize lesion definitions, laterality, diagnostic timing and field provenance into an exposure-feasibility checklist. Registration and clinical professionals then confirm the population and classification design.
How can QSevidence support comparison of location-specific survival evidence?
It can help structure OS, CSS, time origins, competing events, adjustment variables and missing-data methods in an evidence table. Statisticians evaluate associations and uncertainty using original data.
How can QSevidence help evaluate the value of a new location variable?
Teams can compare baseline and extended models, calibration, discrimination and validation designs while preserving source versions. Reproducible analysis and external validation inform professional judgments about utility.
Sources and research notice
Linked sources are official NCI/SEER coding and data documentation and EAU guidance. This article reports no newly extracted SEER findings and supplies no unverified cohort estimates or treatment-effect conclusions. It supports study design and evidence collaboration; qualified clinical teams make individual care decisions using the actual circumstances.