Choosing Group Sizes for Autoimmune Efficacy Studies
You are here: Home » News » 02. Preclinical Research & Model Selection » Choosing Group Sizes for Autoimmune Efficacy Studies

Choosing Group Sizes for Autoimmune Efficacy Studies

Views: 0     Author: Site Editor     Publish Time: 2026-11-16      Origin: Site

Inquire

wechat sharing button
line sharing button
twitter sharing button
facebook sharing button
linkedin sharing button
pinterest sharing button
whatsapp sharing button
sharethis sharing button

HKeyBio Industry Commentary

Choosing Group Sizes for Autoimmune Efficacy Studies

HKeyBio Q&A Series 44

Originally published in Chinese on June 18, 2026.

HKeyBio’s take. How many animals per group give an autoimmune efficacy study enough precision to support a development decision? Start with the question the study needs to answer, the smallest effect that would matter and how variable the primary endpoint is in that model. The same group size may be sufficient for one endpoint and too small for another.

Q1  Why Use a Study Specific Sample Size

Six or ten animals per group are familiar starting points. The number needed for a given study depends on five factors [1,2].

· Variability in the disease model

· Expected treatment effect

· Significance level (α)

· Target statistical power

· Type of endpoint

Animal-to-animal variation can obscure a treatment effect when disease scores fluctuate widely. A reproducible model and a large treatment effect can support a smaller group. The primary endpoint and statistical analysis should reflect the decision the study is designed to support.

Q2  Which Statistical Factors Drive Sample Size

Effect size is often a major driver, particularly in relation to the variability of the endpoint. Consider an arthritis model in which the model group scores 8.0 ± 2.0 and the treatment group scores 4.0 ± 2.0. The treated group mean is 50% lower. If the mean falls from 8.0 to 6.8, it is 15% lower. With the same variability, significance level and target power, detecting the smaller difference generally takes more animals. The percentages compare group means. For sample-size planning, the spread measure represented by “±” must be specified.

A pilot can estimate group means, standard deviations and the likely magnitude of the response. Small pilots yield imprecise estimates, so use comparable historical data where available and check how the required N changes across plausible values. For the main study, specify an effect large enough to matter biologically or to change a development decision [1,2].

Q3  Why Group Sizes Differ Across Disease Models

Induction and progression vary across models and between animals. Dextran sulfate sodium (DSS) colitis, imiquimod (IMQ)-induced psoriasis and ovalbumin (OVA)-based allergy models can show relatively consistent induction under established conditions. For these models, six to eight animals per group is an experience-based starting range.

Collagen-induced arthritis (CIA), systemic lupus erythematosus (SLE) models and humanized immune-system models may show wider variation in onset and severity, even within the same cohort. Some studies in these settings use eight to twelve animals per group or more. These figures reflect experience under particular conditions. When comparing sample-size needs, assess the variability of the chosen endpoint as well as the size of the expected treatment effect.

Q4  Why Some NHP Studies Use Three to Six Animals per Group

Small nonhuman primate (NHP) studies can address focused translational questions. For a pharmacologically relevant species, similarities to humans in immune-cell populations, Fc receptor expression and other aspects of immunity can help interpret responses. A small exploratory cohort can characterize the direction, magnitude and consistency of a response and inform a follow-on study. The precision needed for a confirmatory efficacy claim requires its own justification [1,2].

Apply the 3Rs. Replacement, Reduction and Refinement guide the use of animals. Pairing or matching, repeated measurements and baseline-adjusted analysis can improve precision when the design and endpoint support them. Repeated samples from one animal still represent one experimental unit if the animal is independently assigned to treatment [2,3].

Use early data to plan a separate follow-on cohort, or prespecify any sample-size reassessment within the same study, including the rule for adding animals and how inference will be preserved. Avoid expanding enrollment in response to an interim p-value without a planned analysis.

Q5  Why Statistical Power Matters

Power is the probability that a planned test detects a statistically significant difference when the specified effect is real, under the assumptions used to design the study. A target of 80% means that, if the planned effect size and variability are correct, about 80% of repeated studies under that design would detect that effect at the chosen significance level [1,2].

When power is low for a meaningful effect, a real effect may yield a nonsignificant result. That is a false negative under the assumed scenario. In drug development, treating every nonsignificant result as evidence of no effect can lead a team to drop a promising candidate. The goal is to balance the strength of the evidence with animal use and study cost.

Q6  What Group Sizes Can Guide Initial Planning

The original article gives the following experience-based ranges for discussion. Final group size should reflect the study objective, relevant historical data, pilot results and an analysis appropriate to the endpoint. Allow for expected non-induction or attrition where relevant and report both the numbers assigned and the numbers analyzed [1,2].

Table 1. Illustrative numbers of animals per group from the original article. Use these ranges for initial planning discussions.

image.png

 Q7  How to Meet Statistical Requirements and Control Costs

First, run a small pilot. Estimate the proportion of animals that develop the intended phenotype, disease scores, variability and preliminary treatment response. Report the coefficient of variation where appropriate for the scale of the endpoint.

Second, run a power analysis. Define the primary endpoint and the minimum meaningful effect. Specify α, target power and variability assumptions. Use a statistical method matched to the planned analysis to estimate the required N, and compare scenarios if assumptions remain uncertain.

Third, optimize the design. Improve model consistency, use baseline-stratified randomization when relevant, use repeat measurements if they improve precision, and remove groups that do not answer a study question. Better design can improve efficiency more than increasing group size alone.

A defensible group size gives the team enough information to detect or estimate a meaningful effect for the intended translational decision. Plan around the expected effect, model variability, required power and study objective. This supports more credible conclusions and can reduce repeat experiments and avoidable cost.

Author and reviewer. HKeyBio Marketing Department

Edited by Jing Yan

About HKeyBio

HKeyBio is a preclinical contract research organization providing end-to-end in vivo efficacy services for autoimmune and allergic diseases. Its core technical team has more than 20 years of experience in autoimmune drug development. The company has ISO 9001 certification and operates molecular, cellular, pathology, small-animal and analytical facilities in Suzhou Industrial Park, with nonhuman primate study facilities in Suzhou and Guangxi. The original article reported approximately 80 employees and project leads with master’s or doctoral degrees. The team supports preclinical research and investigational new drug (IND) submissions.

· 50+ nonhuman primate models of autoimmune and allergic diseases

· 500+ animal models of autoimmune and allergic diseases

· Experience supporting 500+ IND submissions in autoimmune disease

References

[1] Percie du Sert N, Hurst V, Ahluwalia A, et al. The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research. PLoS Biol. 2020;18:e3000410. Source

[2] Percie du Sert N, Ahluwalia A, Alam S, et al. Reporting animal research: Explanation and elaboration for the ARRIVE guidelines 2.0. PLoS Biol. 2020;18:e3000411. Source

[3] NC3Rs Experimental Design Assistant. Experimental unit. Source

RELATED NEWS

HkeyBio is a China-based, globally-focused preclinical CRO dedicated exclusively to autoimmune and allergic diseases, with a unified service network across Asia, the Americas, Europe and Australia.

CONTACT US

Phone: +1 2396821165 / +86-512-67485716
Email: tech@hkeybio.com 
Suzhou HQ: 388 Xinping Street, Suzhou Industrial Park, Jiangsu, China 
Shanghai: 489 Yunchuan Road, Baoshan, Shanghai, China 
Nanning: Fazhan Avenue, Xixiangtang, Nanning, Guangxi, China 
Boston: 134 Coolidge Ave, Suite 2, Watertown, MA 02472, USA Melbourne: 432 St Kilda Rd, Melbourne, VIC 3004, Australia

QUICK LINKS

PRODUCTS CATEGORY

SIGN UP FOR OUR NEWSLETTER

Copyright © 2026 HkeyBio. All Rights Reserved.  Sitemap | Privacy Policy