Cell population data: much more to explore

How are CPD calculated?

When a single white blood cell (WBC) passes the detectors in a hematology analyzer, it generates a set of electronic signals, one for each detector that is present. Of course, the absolute value of each signal depends on the cellular characteristics of the WBC in question. An entire blood sample, in which usually 10 to 20,000 WBCs are measured, yields one large dataset (a listmode data file in flow cytometry terms) that is subsequently analyzed by the software for defining clusters of WBCs with similar characteristics. In a normal blood sample 5 or 6 WBC clusters are generally distinguished, which correspond to the usual leukocyte subsets. Once the various WBC clusters are defined, the mean and dispersion of all signals in the WBC populations is calculated and this is called CPD. In principle, the mean and standard deviation of each detector signal can be calculated for each WBC subset, but not all information is accessible, depending on the analyzer manufacturer. A set of CPD can be regarded as the fingerprint of a patient’s WBC at a given moment and CPD harbor information that is directly or indirectly associated with blood cell morphology. For example, neutrophil volume is associated with the absence or presence of immature granulocyte forms, neutrophil side scatter signal (granularity) represents the degree of granulation and granules size and lymphocyte volume can point to an activated status due to viral infection. Currently, all CPD results are labeled ‘for research use only’, not to be reported to physicians.

WBC differential flags and CPD

Once the WBC clusters have been defined, the analyzer’s algorithms try to find anomalies in size and location of the clusters and also to detect overlap between clusters that are normally spatially separated. This process may result in alerts and WBC flags, some of which automatically invalidate the WBC differential results. For example, this can occur when blasts are present (overlap with lymphocytes and/or monocytes) or when large, activated lymphocytes are involved (overlap between lymphocytes and monocytes). As manufacturers algorithms are proprietary, the effect of WBC flagging on CPD is unknown. However, since CPD are raw data and not subject to validation algorithms, most likely CPD are affected by abnormal WBC clusters. For instance, the increased monocyte volume distribution width (MDW), which has been proven to be associated with sepsis and severity of Covid-19 [4], [5], [6], may be in part due to non-monocytic cells located in the cluster designated as monocytes. It remains to be investigated what this means for the clinical utility of CPD in these disease states.

Inter- and intra-analyzer variation in CPD

Apart from being the fingerprint of the patient’s WBC, CPD is also characteristic of the specific analyzer used for measuring the blood sample. Obviously, analyzers with different detection principles will give different CPD. But also analyzers of the same brand and the same type may produce different CDP results. Optical and electronic components used for building the optical bench of a given hematology analyzer have to meet strict specifications with close tolerances, but they can never be produced and adjusted so precisely that they produce exactly identical signals as another analyzer, even of the same type. In the literature published so far, this aspect is widely neglected. To the best of my knowledge, only a single paper has addressed the variability of CPD among multiple analyzers and difficulties of harmonization between analyzers and between laboratories [7]. Because CPD cannot be calibrated, optical and/or electronic adjustments form the only possibility for changing CPD, obviously unreachable for a user.

Some published CPD studies were performed using a single hematology analyzer and then inter-instrument variability does not play a role. Many other studies failed to indicate whether one or multiple analyzers were used for collecting CPD results, which is a drawback from an analytical perspective. Reports on multi-center studies on CPD appear to be rare and none of these address the potential issue of differences between CPD originating from multiple analyzers [8], [9], [10], [11].

Another limitation of studies on the CPD topic is that data on long-term stability of CPD are nonexistent. Most likely CPD are quite stable in a single analyzer, but it is impossible to apply traditional methods of internal QC, also due to the unavailability of suitable control material. If the analyzer allows, moving averages of raw measurement data is probably a good method for monitoring CPD stability, but published evidence is scarce [7]. Therefore, the effect of service interventions, in particular when the analyzer’s optical bench is involved, remains unknown, too.

Pre-analytical effects on CPD

Until present, pre-analytical influences on CPD have hardly been addressed in the literature. There is some evidence that blood storage may affect CPD [12, 13]. Another report indicated CPD stability for 6 h, referring to unpublished results [14]. Although pre-analytical factors represent an almost unexplored topic in CPD research, one could argue that they are nevertheless important. Namely, effects of EDTA anticoagulant on blood cell morphology are well documented and since CPD basically represent cellular morphology, it is very likely that factors like storage time and temperature will affect CPD results to a considerable degree. It is also well conceivable that the technology applied in a hematology analyzer contributes to its susceptibility to pre-analytical effects, meaning that pre-analytical effects have to be investigated for each technology. Many reported CPD studies are retrospective, which carries the risk of ill-controlled pre-analytical conditions, resulting in variable results and potentially incorrect conclusions.

Biological variation of CPD

Nowadays, data on biological variation and reference change values are considered indispensable for clinical laboratory result interpretation, but surprisingly few publications are available on this subject for CPD [15, 16]. In particular when CPD are used for disease severity or risk assessment in an individual patient, this information needs to be available.

CPD reference intervals

In view of all topics discussed above it remains questionable to which extent reference intervals established in healthy subjects have practical usefulness for interpreting patient data [17], [18], [19], [20]. This requires more investigation.

Conclusions and recommendations

Despite the rapidly growing body of evidence demonstrating the clinical usefulness of CPD in a variety of clinical situations, much is still unknown on CPD. Consequently, many analytical and pre-analytical aspects of CPD remain to be explored. Recommendations for future analytical research include inter-analyzer and between-laboratory comparability and harmonization, possible interference by cells belonging to other morphological WBC types, long-term CPD stability within an analyzer and methods for quality control of CPD. Now CPD are no longer interesting “toys” but have evolved into tools with proven clinical utility, the diagnostic industry should support researchers by supplying relevant information, even despite CPD are non-reportable research parameters.

On the pre-analytical side factors like the influence of anticoagulants and time- and temperature-induced changes in WBC morphology with an effect on CPD need to be examined, in order to optimize clinical applicability. So overall there is much more to explore concerning CPD.