Update NAb methods: presentation of statistical endpoints

[brborate]

Location

statistical-methods.Rmd

Old

The template currently states:

Response to a pseudovirus was considered to be positive if the [net] ID~50~ titer was greater than or equal to 20.

Response rates were calculated for each group and time point with corresponding 95% two-sided Wilson score confidence intervals.

The response magnitude was defined as the log~10~ ID~50~ titer.

To measure durability, the fold-change in response magnitude (log~10~ net ID~50~ titer) between week [peak] (peak time point) and week [post-peak] was calculated. The fold-change was only calculated for pseudoviruses with samples at both time points and for animals with positive responses at week [peak], defined as:

$$\mathrm{Fold\mbox{-}change} = \frac{X{\text{week [peak]}}}{X{\text{week [post-peak]}}}$$

where X represents the response magnitude.

New

1. Response magnitude (log~10~ ID~50~ titer).
2. Positive response ([net] ID~50~ titer greater than or equal to 20).
3. Durability: the fold-change in response magnitude (log~10~ net ID~50~ titer) between week [peak] (peak time point) and week [post-peak]. The fold-change was only calculated for pseudoviruses with samples at both time points and for animals with positive responses at week [peak], defined as:

$$\mathrm{Fold\mbox{-}change} = \frac{X{\text{week [peak]}}}{X{\text{week [post-peak]}}}$$

where X represents the response magnitude.

Justification

Statistical Endpoints are critical and it could help if they are laid out point-wise without being verbose. An additional paragraph could follow if details around them need to be explained. Refer Wu001 NAb PT report. https://github.com/FredHutch/Hassell750Analysis/blob/master/NAb/final_report/Nab_unblinded_PT_report.pdf

[asatofh]

Summary from 2/28/24 hackathon discussion with @brborate @tzujunghuang @yungwenliu @asatofh:

• Statistical endpoints section could be more concise.
• • Counter point: we could make the language somewhat more concise, but the goal of this section is not a bulleted list of endpoints, it's a description of the calculations. Language is based off the HVTN NAB memo template, which was approved by the lab.
• Statistical endpoints should contain only those endpoints which involve some derivation (e.g., not response magnitude as written).
• • Counter point: we should add more detail on how the ID50 titer is calculated. This would make it clearer how the magnitude is derived.
• Fold change definition is not standard (should be durability over peak).
• Wilson score CI references could be moved out of endpoints into a different section of the statistical methods section.
• The term "biological endpoints" caused some confusion.

Possible next steps / action items (Alicia's proposal):

• Add additional language to the response magnitude definition. (need volunteer)
• Move sentence about Wilson CI to a new section of statistical methods (response rates? statistical tests?).
• Modify the definition of fold-change to be durability divided by peak. Check whether this affects the description of fold-change later in the statistical methods. (need volunteer)
• Consider renaming "Biological endpoints" to "Endpoints" (similar to HVTN template). Consider renaming "Statistical endpoints" to "Endpoint calculations".

@brborate @tzujunghuang @yungwenliu Please review next steps and self-assign to steps that need a volunteer. Thanks! @at-crx @lemireg @ZaczZhang Please review and let me know if you have additional input. @mayerbry FYI

[asatofh]

@sshotorb FYI

[kelliemac]

Note: Alicia and Solmaz are working on BAMA first and then will tackle this.