Toxicology report review that checks every CRO draft against its own tables, in hours
Every CRO tox report checked against its own tables in hours, with every margin cited.
See one case, screen by screen ↓Software that helps a sponsor check each CRO toxicology report before it is final: the summary against the tables, the NOAEL and its exposure, the margin against the clinical dose. Toxicology Report Review adds six agents that read every draft as it arrives, extract each value with its page, check it against itself, its protocol and the programme, and draft the comment letter — for the study monitor to decide and sign.
How to choose AI solutions for life sciences →
Updated
A 13-week GLP toxicology draft runs 300–500 pages: the summary, methods, clinical pathology and organ weight tables, the pathology report, individual animal listings, the TK report, the protocol and its amendments. The sponsor's study monitor has to read it inside the CRO's comment window, check that the summary says what the tables say, decide which findings are adverse, confirm the NOAEL and send a numbered comment letter. By hand that takes two to four days per draft, and a study monitor may have three drafts open at once.
The errors that matter are small and specific: a mid-dose ALT change left out of the summary, an incidence count that differs from the pathology table, the NOAEL exposure quoted from Day 1 instead of the last sampling day, a dose formulation result out of range with no deviation, a second NOAEL in the discussion. Each one moves the NOAEL, a target organ or the margin — and each one found after the study director signs the final report costs a GLP amendment. Meanwhile the programme team asks “what dose can we go to?”, and the margin spreadsheet is rebuilt by hand for every dose.
Estimated split for one 13-week repeat-dose draft, by hand and with the solution, using the working solution's baseline of 22.4 hours by hand and its 6.4-hour target.
Six agents read, extract, compute margins, check and draft the CRO letter; the toxicologist decides each issue and signs.
Corventa (NWB-417) is an oral small molecule in Phase 2 at 200 mg once daily for 12 weeks, and the programme team is considering 400 mg. Two 13-week drafts are open with their CROs. Here is what happens next, screen by screen, in the working solution.
Organs run down the side, the eight Corventa studies across, each cell the finding in plain words — “ALT ×2.1 · hypertrophy ≥30” in the 13-week rat, “Emesis, soft faeces at 45” in the 13-week dog. Each row ends in that organ's margin at 200 mg: liver 5.1× and GI tract 5.1× from the dog, testis 7.0× from the rat, hERG 49×. Red dots mark the cells the checker flagged; the rat draft has 6 open issues, the dog draft 3.
“Every adverse target organ keeps at least 5× at 200 mg once daily. Lowest: liver 5.1× (CRV-T-014).”
The team wants the margin at 400 mg before the governance meeting. With the clinical dose switched to 400 mg, human AUC goes to 12,900 ng·h/mL and every margin is recalculated: liver falls to 2.4×, GI tract to 2.4×, testis to 3.3×, and their rows turn red with what the margin was. The hERG margin stays at 24×.
“3 of 3 adverse target organs fall below 5× at 400 mg once daily … Highest dose that keeps every adverse organ at 5×: about 205 mg.”
Fourteen CRO reports across three programmes, with the stage, discrepancies, NOAEL and the comment due date for each. Four are ready for her. CRV-T-013, the 13-week oral toxicity study in Wistar Han rats from Ashgrove Preclinical, is Draft v1 at 412 pages: 5 major and 1 minor discrepancy, NOAEL 30 mg/kg/day, comments due Oct 16.
“Working copies only — the signed GLP report stays the official record.”
The report reader splits 412 pages into sections and tables and matches them to the protocol and amendments in 41 s. The findings extractor pulls 214 findings, 7 of them test article-related, with SEND terms mapped. The exposure and margin calculator reads the TK tables, the consistency checker runs its 7 checks and finds 6 discrepancies, and the cross-study check compares the draft with the 7 other Corventa studies and flags the thyroid.
Read, extracted and checked by 6 agents in 3 min 14 s — the last step is “Study monitor decides”.
Centrilobular hepatocellular hypertrophy in 6/10 males and 4/10 females at 30 mg/kg/day and 10/10 at 100; ALT +46% at 30 and 2.1-fold at 100 on Day 91. Each finding keeps the CRO's words beside the SEND term, a severity, a call and its recovery, and a number that opens the cited page. The call is the report's own — “Adaptive at 30 · adverse at 100 with ALT” — and the toxicologic pathologist confirms it before the review is signed.
The summary on page 6 gives the Day 91 AUC at the NOAEL as 48,200 ng·h/mL in males. TK Table 3 on page 389 shows Day 91 male AUC at 30 mg/kg/day of 42,800 — 48,200 is the Day 1 value. Compare puts both pages side by side with the values highlighted, and the checker drafts the comment to the CRO.
“Exposure margin at 200 mg falls from 7.9× to 7.0× with the correct value.”
The rat margin at the NOAEL is 7.0× at 200 mg, on males as the lower-exposure sex; females at 61,500 ng·h/mL give 10×. The human-equivalent dose is 290 mg — 30 mg/kg ÷ 6.2 for the rat, by body-surface area, for a 60 kg adult. A 13-week rat study supports clinical trials up to 3 months under ICH M3(R2); the plan is 200 mg once daily for 12 weeks.
The calculator uses the lower-exposure sex unless told otherwise, and states the basis of every margin.
Priya decides each issue: the ALT change missing from the summary, testis degeneration in 1/10 males in the summary but 2/10 in Table 22 (animals 4107 and 4112), the Day 1 TK value, the Week 6 high-dose formulation at 87.0% of nominal with no deviation recorded, thyroid missing from the target organs, and an organ weight header in the wrong unit. All six go to the letter for Dr. Elin Varga at Ashgrove, one numbered comment each with its location and class.
“Please reply within 10 working days, by Oct 21. The letter is sent from the CRO portal after your signature.”
The signing screen shows what she is signing: 6 comments to Ashgrove, and that signing confirms the remaining 7 findings as extracted. She chooses the meaning — “Reviewed — comments sent to the CRO” — and signs with her password as study monitor, Toxicology.
“Your signature, the meaning, the time and report Draft v1 are recorded in the audit trail.”
Only after her signature does the CRO letter drafter deliver the letter to Ashgrove Preclinical through the CRO portal; then the Corventa cross-study table is refreshed. The activity trail shows every step with its time — the agents' reading on Oct 2, her decisions, her signature and the letter on Oct 7. The next draft will be re-checked against every item automatically.
“Signed · 6 comments sent to Ashgrove Preclinical · cross-study table refreshed”
The same draft and the same programme, seen by the five people who carry them — what their week looked like, and what it looks like now.
Six agents read, extract, compute margins, check and draft the CRO letter; the toxicologist decides each issue and signs.
Splits each CRO report into summary, methods, results, tables, listings, pathology, TK and appendices, keeps page numbers and matches the report to its protocol, amendments and earlier versions.
Extracts dose groups, findings by organ with SEND terms, incidence by sex and dose, severity, reversibility and the stated NOAEL — each value with its page.
Reads the TK tables, picks the right day, sex and dose, and computes exposure margins and human-equivalent doses against the clinical exposure reference.
Checks the summary against the tables, the tables against individual listings, the TK report against the summary, the report against its protocol and amendments, and the study against the rest of the programme.
Turns the accepted discrepancies into a numbered comment table for the CRO study director, with the location, what is wrong and what the sponsor asks.
Keeps the programme's cross-study table and target-organ map up to date — target organs, NOAELs, reversibility, exposure and margins, every cell cited — and recomputes the margins when the clinical dose changes.
Decides each issue and signs. The agents propose; a named person decides.
Ask about any CRO report, finding, NOAEL or exposure margin in your programmes. Every answer points to the cited page.
What is our Corventa margin at 400 mg?
At 400 mg once daily (human AUC 12,900 ng·h/mL) the dog sets the lowest margins — 2.3× in the 4-week study and 2.4× in the 13-week study; the 13-week rat falls to 3.3× using the Day 91 value from TK Table 3. The hERG margin stays above 20× against free Cmax. Liver, GI tract and testis fall below the 5× floor. The highest dose that keeps every adverse organ at 5× is about 205 mg.
What did the checker find in CRV-T-013?
Six discrepancies in the 13-week rat draft, five of them major: the ALT change at 30 mg/kg/day missing from the summary; testis degeneration 1/10 in the summary, 2/10 in the tables; the NOAEL exposure quoted from Day 1; the Week 6 high-dose formulation at 87% with no deviation recorded; thyroid missing from the target-organ list; and an organ weight header that says “% body weight” over values in grams. The TK one changes the margin at 200 mg from 7.9× to 7.0×.
Which target organs are consistent across species?
For Corventa, liver is the only target organ in both species — hypertrophy with ALT up to 2.1-fold in the 13-week rat and ALT up to 1.8-fold in the 13-week dog. Thyroid follicular hypertrophy is rat-only and adaptive (liver enzyme induction). Testis degeneration is rat-only, at 100 mg/kg/day, and not fully reversible. GI intolerance is dog-only.
How many males had testis findings?
Two high-dose males in CRV-T-013, not one: Table 22 lists 2/10 at 100 mg/kg/day, and the individual animal listing names animals 4107 (unilateral) and 4112 (bilateral), both minimal. The summary says 1/10 — that is discrepancy D2. One recovery male still had the finding, so it is not fully reversible.
13 screens from the working solution, on its sample data. Pick one to see it large.
What the reviewer checks, the clinical exposure it compares against, and who signs are settings, on the audit trail.
| Setting | Default | Choose from |
|---|---|---|
| Checks the reviewer runs | 7 on | Summary against tables · incidence against individual listings · TK report against the summary · report against protocol and amendments · dose formulation and deviations · target organs across the programme · terminology against SEND — each on or off |
| Severity of a summary-vs-table mismatch | Major | Major · Minor |
| Margin floor | 5× at the NOAEL | 3× · 5× · 10× at the NOAEL |
| Studies that set an organ's margin | As long as the clinical plan | As long as the clinical plan · All repeat-dose studies |
| Exposure basis | Lower-exposure sex · AUC | Lower-exposure sex · AUC · Mean of both sexes · AUC · Cmax |
| CRO comment window | 10 days (Ashgrove Preclinical) | 10 · 15 · 20 days, per CRO |
| SEND controlled terminology | 2026-06 release | 2026-06 release · 2026-03 release |
| Severity grading scale | 5-grade (minimal–severe) | 5-grade (minimal–severe) · 4-grade |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample programme data · “target” = the design goal, measured in the live solution · “estimated” = our estimate (by-hand review of 2–4 days per draft is common practice, not a published figure). Sources: ICH M3(R2) (a 13-week repeat-dose study supports clinical dosing up to 3 months) · 21 CFR 58.185 (the GLP final report, and amendments to it). People, companies, CROs, programmes and studies named on this page are fictional — characters and sample data in the working solution.
Software that helps a sponsor check each CRO toxicology report before it is final: the summary against the tables, the NOAEL and its exposure, the margin against the clinical dose. Toxicology Report Review adds six agents that read every draft as it arrives, extract each value with its page, check it against itself, its protocol and the programme, and draft the comment letter — for the study monitor to decide and sign.
The findings extractor reads the stated NOAEL, findings by organ, incidence by dose and sex, severity and reversibility, each with its page. The exposure and margin calculator reads the TK tables and picks the last sampling day and the lower-exposure sex unless told otherwise, so a summary that quotes the Day 1 value is caught.
Margin is the animal exposure at the NOAEL divided by the human exposure at the chosen clinical dose, from the clinical exposure reference in Settings; heart and QT uses the hERG IC50 against free Cmax. By default only repeat-dose studies at least as long as the clinical plan set an organ's margin, and the basis of every margin is stated. Change the dose and every margin on the map is recalculated.
Summary against tables, incidence counts against individual listings, TK values, unreported protocol deviations such as an out-of-range dose formulation, target organs inconsistent with the programme's other studies, and labels and terminology against SEND. Each discrepancy shows both sources and why it matters.
No. The extractor marks adverse or adaptive only as the report states it, the consistency checker proposes and never decides, and the toxicologic pathologist confirms adverse versus adaptive calls before the review is signed.
Yes. The study monitor accepts or dismisses each discrepancy, with a reason on the record for a dismissal, and signs the sponsor review. The comment letter leaves through the CRO portal only after that signature, and no agent changes a source report.
Yes. The cross-study table is kept up to date from confirmed findings, every value cited to its report page, with draft values labelled “may change”. It exports for the IB and CTD Module 2.6.7 with source references, so quality assurance can QC each value against the signed GLP report.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your review SOP, clinical exposure reference and a sample of past CRO reports — and it goes live once every quality gate has passed. We will walk you through it on your own reports first.
Sourced answers from papers, patents and reports in seconds, our own disagreements first.
1-min demoCompetitive Pipeline TrackingEvery rival trial, readout and approval in a sourced brief each Monday
1-min demoPreclinical Study Report AutomationA draft GLP report in hours, every number traced to raw data.
1-min demoWe'll run Toxicology Report Review on a sample of your own past CRO draft reports.