Question: Is RNA quality assessment (either by RIN or DV200) important for Xenium?
Answer: As a probe based assay, Xenium is robust to RNA degradation often seen in FFPE blocks. Xenium works across a wide variety of samples and do not observe a strong correlation with DV200 or RIN score (DV200 >= 30). Running samples with DV200 score below 30% potentially risks decrease in assay performance as measured by lower median transcripts per cell. Here is a summary of the lessons learned and why sample QC for Xenium is multifactorial.
An Overview of DV200
DV200 is the percentage of RNA fragments >200 nucleotides. It is important to note that while DV200 are predictive of the average RNA quality of the section surveyed, DV200 scores may vary within a block.
For example, the outer layer may be exposed to more environmental factors. Also, tissues that are not homogenous, may have different tissue quality depending on the region surveyed or disease pathology. In general, DV200 score above 30% is considered a good RNA quality sample Visium CytAssist Recommendations and Illumina Tech Note on evaluating FFPE sample quality).
Evaluation of Xenium Sample Inputs
For Xenium, assay performance (as measured by transcripts per cell, decoded transcript density, and transcript quality scores) depends on many factors including:
- The sample itself:
- Disease state and patient age impact gene expression.
- Tissue to panel match:
- A well designed panel can, on average, increase the number of transcripts detected per cell (irrespective of DV200) as shown by the Alzheimer samples below.
- Sample fixation method impacts both RNA quality and accessibility:
- Sample storage conditions:
- UV damage and heat exposure are suboptimal for RNA-based assays.
- How should I store my FFPE tissue blocks?
- RNA integrity.
- In general, Xenium is robust to RNA degradation often seen in FFPE blocks. In validation experiments, blocks with DV200 scores from 18 to 70 were tested. We observe robust decoding performance on a variety of DV200 scores as exemplified by the fraction of rolling circle products (RCPs) being approximately the same across all DV200 scores tested, demonstrating the ability to generate amplicons even if RNA is fragmented.
- The median transcript per cell varied significantly depending on the sample/block used. This seems more related to tissue to panel match, tissue type, disease state, and presence of add-on panel (as described above). The addition of an add-on panel increased the median transcripts per cell. Also certain sample types, like cancers, had fewer median transcripts even with higher DV200 scores.
In summary, we observe high quality decoding efficiency, as defined as ~80% of transcripts decoded have a quality score greater than or equal to 20, across a broad range of DV200 scores. These high confidence transcripts can be used for downstream analysis. We expect samples that have been shown to be successful with Flex or Visium to be successful with Xenium.
Despite our observation of a low correlation of DV200 score with decoding efficiency, running samples with DV200 score below 30% potentially risks decrease in assay performance as measured by lower median transcripts per cell.
Troubleshooting Low Median Transcripts Per Cell
If a sample has low decoding or transcripts per cell, we would recommend a holistic approach to troubleshooting, weighing the following:
- Tissue to panel match (as assessed by reference based assessment on matched datasets)
- Fixation conditions of the sample relative to the best practices outlined above.
- RNA fragments greater than 40 bp for probe binding.
- Tissue morphology by H&E assessment. Details for this can be found in the Xenium Tissue Preparation Guides.
Product: Xenium In Situ Gene Expression