Most nonclinical teams at mid-sized biotechs discover their SEND dataset problems at the worst possible moment: weeks before an IND or NDA filing deadline, when a quality review reveals that the submitted data cannot reproduce the summary tables in the signed study report. PointCross Life Sciences has processed over 10000 studies, and the pattern is consistent. The datasets pass validation and the conformance checks come back clean, but the numbers in the SEND package and the numbers in the GLP study report do not agree.
That gap between passing validation and actually being reviewable is where regulatory submission delays originate. And it is precisely where the evaluation criteria for a SEND dataset service provider should begin.
Why does validation alone fail to catch submission-blocking errors?
Conformance validation confirms that your SEND datasets follow CDISC SENDIG structure and controlled terminology rules. It checks variable lengths, required fields, cross-domain references, and data types. What it does not check is whether the values in those fields match the audited study report.
A body weight mean in the BW domain can be correctly formatted, correctly coded, and correctly placed in the .xpt file, yet still disagree with the mean reported in the study reportβs summary table. An incidence count in the MI domain can pass every CDISC business rule and still not match the pathologistβs tabulation. These discrepancies are invisible to standard validation engines.
The FDA has made its position on this increasingly explicit. The Study Data Technical Conformance Guide v6.2, released in 2026, added Appendix I: SEND Data Review Prior to Submission. That appendix formalizes three categories of submission failures that prevent efficient regulatory review: submission-level structural issues (incorrect folder placement, missing files), SENDIG conformance issues (empty variables, broken references), and study design representation issues (incorrect group definitions, conflicting results, inconsistent summaries across datasets and reports).
The third category is the one that validation alone cannot address. It requires that your SEND datasets accurately represent the underlying nonclinical study, not just conform to the standardβs technical rules.
What does study report alignment actually require?
Study report alignment means that every summary tabulation published in the signed GLP study report can be regenerated from the SEND dataset. Body weight means and standard deviations in the BW domain must match the report. Clinical pathology group means in the LB domain must match. Microscopic finding incidence counts in the MI domain must match the pathologistβs tables.
This is a 100% verification exercise, not a sampling exercise. Spot-checking ten percent of values and extrapolating misses systematic errors that affect entire domains. When a SEND team manually transcribes source data into dataset domains separately from the team that built the study report, discrepancies are not random. They follow patterns tied to mapping decisions, rounding logic, or grouping variable assignments.
The FDA webcast from CDISC earlier this year delivered an unusually direct message on this point. For the first time publicly, FDA representatives acknowledged that they are rejecting SEND datasets because the quality is not good enough, not merely failing them on technical rejection criteria. The agency described two sides of the quality problem: compliance with the standard (variable lengths, controlled terminology) and trustworthiness of the data (whether SEND accurately reflects the study report).
When you evaluate a SEND dataset service, ask a specific question: do you verify every summary table, or do you spot-check a sample? The answer tells you whether you are buying a conformance check or actual submission readiness.
How should you evaluate synchronization between study reports and SEND datasets?
The traditional model treats SEND dataset creation as a sequential step that begins after the study report is finalized. A toxicology study completes, the study director signs the GLP report, and then a separate team takes source data and maps it into SEND domains. This sequencing adds weeks to the submission timeline and forces an after-the-fact reconciliation between two deliverables that were built independently.
That reconciliation is where errors accumulate. The report team and the SEND team work from the same source systems (LIMS, pathology databases, bioanalytical platforms), but they make independent mapping decisions. Grouping variables may be assigned differently. Timing variables for recovery animals may be coded inconsistently. The result is a dataset that technically represents the study, but does not match the report that FDA reviewers treat as the trusted reference.
Look for providers that generate both deliverables from a single governed data model. When the study report and the SEND dataset share the same source, alignment is built into the process rather than verified after the fact. PointCross Life Sciences uses this approach through its Single Track processing model, which produces the GLP study report and the Submission-Ready SEND Datasets simultaneously. The result is study reports delivered approximately two weeks faster than traditional dual-track workflows, with the SEND package available at the time of report generation.
The evaluation question here is structural: does the provider build SEND after the report, or does it build both from one source at the same time?
What compliance gaps should you test for beyond conformance rules?
SENDIG conformance rules cover structural and terminological compliance. They do not cover several categories of error that the FDA has flagged as common causes of review delay.
Controlled terminology version alignment is one. CDISC publishes quarterly updates to SEND Controlled Terminology, adding new terms, modifying submission values, and retiring others. Your Define-XML must reference the exact terminology version used for each study. A test code valid in one quarter may be deprecated the next. When you run multiple studies with different start dates, each one may require a different SENDIG version and terminology release. Ask your provider how they manage version-controlled metadata across concurrent studies.
nSDRG completeness is another. The nonclinical Study Data Reviewerβs Guide explains your SEND datasets to FDA reviewers. Under TCG v6.2, the FDA now explicitly discourages generic boilerplate statements in the nSDRG. Sponsors must document the organization responsible for SEND dataset creation, the software and systems used during generation, validation and verification procedures performed, any unresolved anomalies, and scientific rationale for unresolved discrepancies. An nSDRG that contradicts the Define-XML or omits study-specific decisions triggers reviewer questions and delays.
SENDIG-Genetox v1.0 readiness is a newer requirement. As of March 15, 2025, all in vivo micronucleus and comet assay studies submitted to CDER or CBER must comply with SENDIG-Genetox v1.0, regardless of GLP status or whether the genetox endpoints are standalone or integrated into a broader toxicology study. Ask your provider whether they support SENDIG-Genetox and whether their validation engines include the associated rule sets.
How do FDA technical rejection criteria affect your choice of provider?
The FDA enforces Technical Rejection Criteria (TRC) through automated checks at the electronic submissions gateway. A high-severity failure means your submission is bounced before a reviewer opens the file.
Four validation rules carry high severity for nonclinical submissions. Rule 1734 requires a Trial Summary dataset (ts.xpt) with a valid study start date. Rule 1735 checks that Study Tagging File entries correctly reference datasets and Define-XML. Rule 1736 confirms that required dataset files (including dm.xpt and Define-XML) are present in Module 4. Rule 1789 verifies that all study files referenced in the Study Tagging File actually exist.
FDA data from 2023 show that 453 nonclinical studies failed Rule 1734 between September 2021 and February 2023, with 73% of those failures caused by a missing ts.xpt file. Repeat-dose toxicology studies accounted for 71% of failures.
These are structural errors, not scientific ones. A misformatted ISO 8601 date, a missing Trial Summary file, or a study ID mismatch will bounce an entire submission. Your provider should guarantee TRC compliance and back that guarantee with warranty terms that cover remediation at no additional cost.
PointCross Life Sciences validates every SEND package against CDISC Conformance Rules, FDA Business Rules, and PMDA validation rules using eDataValidator, and guarantees zero TRC-based rejections.
What questions should you ask about turnaround time and pricing?
Submission deadlines do not accommodate SEND delays. When a provider quotes a turnaround time, ask what that number assumes about the state of your source data. A five-week estimate that requires clean, pre-formatted LIMS exports is a different commitment than a five-week estimate that accepts raw PDF study reports with handwritten annotations.
Source data flexibility matters. Your provider should accept data from LIMS, pathology systems, bioanalytical platforms, EDC databases, spreadsheets, and PDF study reports. Providers that require a single pre-formatted input type create bottlenecks when source data arrives in an unexpected format, which happens more often than most study teams plan for.
Pricing structure matters equally. Variable pricing based on study complexity creates budget uncertainty that compounds across a submission with multiple studies. Firm fixed-price quotes with guaranteed turnaround times let you plan your submission timeline and budget with precision. PointCross Life Sciences issues indicative quotes within 24 hours and provides no-obligation estimates before any engagement begins.
The evaluation criterion is whether the quoted turnaround and price reflect the actual condition of your data, not an idealized scenario.
How does the evolving SEND landscape affect your provider selection today?
SEND 4.0 is expected to add approximately eight new domains covering immunogenicity, ophthalmology, nervous system assessments, skin tests, cell phenotyping, pharmacokinetic inputs, and additional in vivo genetic toxicology endpoints. The FDA is also evaluating CDISCβs Dataset-JSON format as a potential replacement for the longstanding SAS XPT transport files.
Neither change is imminent as a regulatory mandate. But both signal that SEND requirements will continue to expand. The provider you select today should have the infrastructure to absorb new SENDIG versions and format changes without requiring you to rebuild your submission workflow each time.
Organizations that rely on manual mapping processes and ad-hoc spreadsheet tracking will face rework with every new requirement. Organizations that build on version-controlled metadata repositories and automated data pipelines will absorb these changes as configuration updates. Ask your provider how they handled the transition from SENDIG v3.1 to v3.1.1, and how they onboarded SENDIG-Genetox v1.0. Past adaptability is the strongest predictor of future readiness.
PointCross Life Sciences supports this through the Xbiom platform, which provides a metadata repository and automated standardization workflows designed to adapt as CDISC and FDA requirements evolve.
The regulatory bar for SEND is moving in one direction. The FDA now maintains a library of SEND datasets for advanced analysis, querying toxicity profiles across pharmacologic classes and comparing species responses across product types. Your SEND datasets are no longer disposable submission artifacts. They are permanent analytical assets in the agencyβs review infrastructure. Choosing a service provider that treats SEND as a checked box rather than a quality deliverable creates risk that compounds with every study in your portfolio.
If you are evaluating SEND dataset services for upcoming nonclinical submissions, reach out to PointCross Life Sciences at [emailΒ protected] to schedule a working session or request a no-obligation indicative quote.