Setting the gold standard: Commentary on designing and optimizing high-parameter flow cytometry panels

Abstract

Technological advances in flow cytometry have greatly expanded its capabilities. These have occurred gradually over time, but there have also been several key advances that have more markedly affected the technology and how it is used. A prime feature of flow cytometry is the ability to characterize cell marker expression at the single-cell level as high-throughput and high volume. The number of cell markers that could be measured simultaneously was initially very low but has been increasing almost exponentially over time. There are many reasons for this increase including advances in hardware for the instrumentation, introduction of new types of fluorescent dyes beyond those found in nature, and also advances in analysis tools that not only enable efficient data analysis but have elegantly allowed for new insights into fluorescent dye/detector interactions that provide theoretical bases for optimal staining panel design in high dimensions.

OMIP-102 [1] published here represents a milestone in flow cytometry technology and makes use of and expands upon all of those cumulative advances. Simply the demonstration of a “50-color” staining panel is remarkable, but in addition, the approach to the design and the careful and methodical description of the design process both in the main text and the online material provide a definitive syllabus for staining panel design integrating all of the best practices to date.

Because of the wide breadth of information building upon so many of these major advances, it is worthwhile to break down some of those advances into digestible pieces to highlight the significance of this achievement. One key advance in hardware has been the optimization of the instrument optics to enable the relatively weak fluorescent signal to be subdivided most efficiently across large arrays of detectors. Instrument manufacturers have developed their own methods to achieve this goal and the results have been successful with current routine capabilities to detect separate signals from up to 28 fluorescent dyes, and this is expanding. There are multiple Optimized Multicolor Immunofluorescence Panel (OMIP) publications demonstrating successful staining panels at this scale. Likely the most significant advance representing a new paradigm is spectral optics. This is a brilliant concept that in retrospect seems so obvious as the likely best approach. While full spectrum cytometry was first demonstrated in 2004 [2, 3], it required further hardware and software advances to enable routine implementation by a wide user base.

The potential of exploiting the light spectrum more completely for interrogating fluorescently labeled biological specimens directly called for the development of new fluorescent dyes in order to make high-parameter flow cytometry a reality. As the discovery of natural fluorochromes with the appropriate brightness and spectral characteristics was limiting, luckily, custom-designed dyes became available—examples are the quantum dots [4] and organic polymers [5, 6] (and their tandem combinations with more conventional dyes), both of which stemmed from discoveries that earned Nobel prizes. Those of us designing staining panels can all remember the sudden shift in capabilities brought about by the introduction of these bright dyes.

Perhaps not receiving the same recognition for their game-changing impact on flow cytometry are several alternative approaches to data analysis and representation that have enabled the extraction of meaningful insights from increasingly complex datasets. The introduction of bi-exponential scaling [7, 8] may now seem trivial, just another tool that is universally used, but this “tool” does not simply provide a more aesthetically pleasing data representation. Rather, it is necessary to properly interpret results and discern artifacts affecting the negative space. As dimensionality increased, the potential for hidden artifacts also increased, and panel design became more complicated. Purely empirical design was no longer feasible. It became apparent that “spillover spreading” was important to take into account, and the introduction of a metric to measure this and display in the format of the spillover spreading matrix [9] became an integral tool for panel design. Other types of metrics, such as the complexity and similarity indices, have been developed more recently for full spectrum cytometry that are equally effective in streamlining panel design based on theoretical considerations.

The authors of OMIP-102 [1], published in this issue, showcase a 50-color OMIP, where the thoughtful and methodical selection of marker-fluorescence combinations clearly contribute to the panel's effectiveness in analyzing and visualizing the expression of cellular components. After exploiting all existing tools to aid in this endeavor, they also developed a new metric, the unmixing spreading error, which elucidates how the complexity of the spectral unmixing matrix impacts the rise in noise across all measured cells per fluorochrome, enabling the systematic design of such a high-parameter immunofluorescence panel.

This exemplifies the ideal approach to panel design and optimization. While not everyone may need to create new tools and techniques, when confronted with novel challenges stemming from expanding possibilities that stretch the capabilities of current tools, it is essential to adapt our procedures and embrace innovative thinking.

Stephen C. De Rosa: Conceptualization; writing – original draft. Yolanda D. Mahnke: Conceptualization; writing – original draft.

The authors declare no conflict of interest.