Reproducibility and stability of the immature platelet fraction using Sysmex XN-10

Abstract Background The Immature Platelet Fraction (IPF) is an indicator of thrombopoiesis which is a useful parameter in thrombocytopenia. It demonstrates compensatory mechanisms in production of platelets, but currently not implemented in routine clinical practice. The aim of this study was to establish the reproducibility and stability of IPF, for both percentage (%-IPF) and absolute (A-IPF) measurements. Material/methods: A total of 71 samples, of which 45 for reproducibility and 26 for stability analysis, were assayed for full blood count using the Sysmex XN-10 analyser at room temperature (RT:19–25 °C). For reproducibility analysis, IPF measurements were analysed 11 times by different appraisers using the same sample, while for stability analysis, IPF was measured over fourteen hourly-intervals up to 24 h (n = 21) and then separately extended beyond the point of stability to 72 h (n = 5). Results Reproducibility analysis of %-IPF and A-IPF (n = 45) showed very reliable results, with the range of mean CV% values between 1.25–8.90% and 1.70–9.96%, respectively. On the other hand, overall, stability analysis of %-IPF and A-IPF (n = 21) at RT over 24 h showed reliable results, with pooled mean CV% values of 1.32% and 1.43%, respectively, with no significant difference between %-IPF and A-IPF (p = 0.767 and p = 0.821). All %-IPF and A-IPF values had exceeded the set acceptance criterion of stability (CV% ≥ 10.0%) before 72 h. Conclusions Overall, %-IPF and A-IPF reproducibility and storage at RT for 24 h predominantly demonstrates the suitability of their usage for testing on the Sysmex XN-series analysers.


Introduction
The biological sample stability is the ability to retain the initial result, for a certain time period, within a specified range when stored under a given condition [1] and precision (a concept of variation around a central value or closeness of agreement between results) is to contrast repeat measurements obtained by using the same blood sample and analytic platform [2].Under changed conditions (i.e.calibrators, reagents, laboratory, storage time and/or operators), it is considered to be reproducibility [3].It has been demonstrated that the time interval between collection and analysis and the storage temperature may influence the stability of full blood count (FBC) parameters [4].Moreover, the analytical variability of the haematological platform may also influence the interpretation of the FBC [5].In addition to the traditional platelet count (PC), there are extended platelet function parameters related to morphology and thrombopoiesis available on the latest Sysmex XN-10 automated haematology analyser (Sysmex Corporation), which include the: Mean Platelet Volume (MPV), Platelet Distribution Width (PDW), Platelet Large Cell Ratio (P-LCR), Plateletcrit (PCT), and Immature Platelet Fraction (IPF) [6], both absolute (A-IPF) and percentage (%-IPF) measurements hereby referred to as A-IPF and %-IPF, respectively.To obtain the A-IPF, %-IPF is multiplied by the PC and then divided by 100 [7].
IPF is an analytic research parameter which stains residual platelet mRNA and nucleic acid rich organelles such as rough-surfaced endoplasmic reticulum (rRNA) and mitochondria (mtDNA).It quantifies immature platelets (also known as reticulated platelets) to estimate the rate of thrombopoiesis.The benefit of this estimation is the ability to determine the causes of different thrombocytopenias.
The objective of this study was to determine the stability and reproducibility of IPF, both %-IPF and A-IPF, available on Sysmex XN-10 during storage at room temperature (RT) (19-25 °C) and to explore the time points at which %-IPF and A-IPF exceeds stability.This has previously been demonstrated with conflicting results albeit in studies utilising the Sysmex XE-2100 (Sysmex, Kobe, Japan) [8][9][10][11] and the Sysmex XE 5000 (Sysmex, Kobe, Japan) [12], both of which employ a mixture of polymethine and oxazine, in a method which Sysmex Corporation has subsequently changed in the newer XN-series haematology analysers used in the current study [13].The analytical performance is shown to be heterogeneous for IPF using these different blood platelets; thrombocytes and primary haemostasis; thrombocytopenia; blood cell count; pathology haematological analysers [14].A recent study that focused on comparing IPF in EDTA and citrate samples using the XN-series had simultaneously reported the stability, but with a different study design [15].
To our knowledge, no study to date has been conducted that explores the time points at which %-IPF and A-IPF exceeds stability with a focus on the reproducibility and stability of these measurements using the Sysmex XN-10 automated haematology analyser (Sysmex Corporation).

Study design, analytic method and population
This study was conducted at Royal London Hospital and its partner site Homerton Healthcare NHS Foundation Trust which is the first NHS Trust to install the equipment Sysmex XN-10 automated haematology analyser (Sysmex Corporation) in the UK.This equipment was used to analyse the venous blood samples (EDTA K2 Vacutainer tubes) and it is the equipment with a platelet analysis channel (PLT-F) that utilises a novel fluorescent dye and allows the quantification of the %-IPF and the calculation of A-IPF measurements.The equipment had passed its daily internal quality controls and weekly precision checks and had its performance evaluated.The samples analysed in this study were a total of 71 residual surplus venous blood samples left over after routine testing of each patient and the study was approved by the National Research Ethics Service (NRES).
Reproducibility of the %-IPF and A-IPF measurements was conducted over a two-week period on forty-five venous blood samples (EDTA K2 Vacutainer tubes) by different appraisers with the same batch of reagents using 11 replicate analyses (performed by two operators in a format of 5 and then 6 analyses by operators 1 and 2, respectively).These forty-five patients from whom the samples were taken had different PC (17-675 × 10 9 /L; mean 245 × 10 9 /L), ages (17-76 years; mean 46.6 years) and sexes (20 females and 25 males; females to males ratio of 4:5).
Furthermore, for stability analysis, twenty-one venous blood samples (EDTA K2 Vacutainer tubes) were analysed within 10 min of receipt at 14 different time intervals to assess changes with time.These twenty-one samples were taken from patients with different platelet counts (54-1564 × 10 9 /L: mean 293.4 × 10 9 /L), ages (17-85 years; mean 56.1 years) and sexes (14 females and 7 males; females to males ratio of 2:1).There were thirteen separate hourly measurements recorded at the time of receipt (T0) to twelfth hour (T12) -correlating to the time intervals 1 to 13, followed by a single measurement at 24 h (T24) -correlating to the time interval 14, while all samples were stored at the RT (19-25 °C) throughout the analysis.Since there was limited sample volume, samples were not analysed multiple times in each time interval (T0 to T24 or 1 to 14).
Additional research was conducted to study the time points at which %-IPF and A-IPF exceeds stability.Five venous blood samples (EDTA K2 Vacutainer tubes) were analysed within 10 min of receipt at 13 different time intervals to assess changes with time.The normal PC was in the range of 150 to 400 x 10 9 /L and five samples were taken from patients with different PC (45-947 × 10 9 /L), including two samples with thrombocytosis (PC = 947 and 499 × 10 9 /L), two with thrombocytopenia (PC = 45 and 135 × 10 9 /L) and one sample with a PC within normal range (257 × 10 9 /L).There were thirteen separate measurements at every six hour interval recorded at the time of receipt (T0) to seventy-two hours (T72) -correlating to the time intervals 1 to 13 (i.e.T0, T6, T12, T18, T24, T30, T36, T42, T48, T54, T60, T66 and T72), while all samples were stored at the RT (19-25 °C) throughout the analysis.

Statistical analysis
Due to normal distribution, the mean and standard deviation (SD) values for the reproducibility and stability from the 11 repeated values and the values at T0 to T24/T0 to T72, respectively, were calculated.A measure of precision, percentage coefficient of variation (CV%) was calculated for both the %-IPF and A-IPF measurements and the acceptance criterion was set less than 10%, a decision limit used by majority of the laboratories [16].Statistical significance was set at a p value of ≤ 0.05.All the results and statistical analysis were tabulated and carried out using Excel 2013.

Reproducibility analysis
The data for reproducibility (n = 45; 11 replicates for each sample) performed by different appraisers was collected and the mean value was calculated for PC and separately the mean, SD and CV% values for %-IPF and A-IPF (Supplementary Table 1).For %-IPF, the CV% values were between 1.25-8.90%(with a CV% of ≤ 5% in most samples, n = 40) and demonstrated an excellent reproducibility of the test.Only five CV% values exceeded 5%, but none were greater than 10% -the acceptance criterion set for this study that met the manufacturer's specifications for precision.In contrast, for A-IPF, the CV% values were between 1.70-9.96%(with a CV% of ≤ 5% in most samples, n = 36) and demonstrated a very good reproducibility of the test.In this case, only nine CV% values exceeded 5%, and no sample had a CV% greater than the acceptance criterion of 10%.Overall, both %-IPF and A-IPF showed a very reliable reproducibility using the Sysmex XN-10 analyser.

Stability analysis
The data for stability (n = 21; T0-T24) was collected with age, sex, and primary diagnosis that was available for 17 out of 21 patient samples.The SD and CV% values for each sample were calculated for both %-IPF and A-IPF and the values predominantly met the set acceptance criterion (CV% < 10%) for good stability over 24 h.However, 3 out of 21 samples (i.e.samples 2 to 4) unexpectedly exceeded the acceptance criterion (CV% < 10%) for both %-IPF and A-IPF, for samples taken from patients with a primary diagnosis of acute liver disease (sample 2), undefined (sample 3) and sickle cell disease (sample 4) (Supplementary Table 2).
Figures 1 and 2 graphically illustrate the pooled mean values (n = 21) for %-IPF and A-IPF, respectively, over 14 time intervals, on an hourly basis, at the time of receipt (T0) to the twelfth hour (T12), and then again at the 24 th h (T24).For %-IPF, the pooled mean SD and CV% was 0.08 and 1.32%, respectively, illustrating an overall excellent stability of the test over a 24 h period.There was no statistically significant difference between %-IPF values obtained at T0 and T24 (p = 0.767).For A-IPF, the pooled mean SD and CV% were 0.21 and 1.43%, respectively, which is also an overall excellent demonstration of the test stability over a 24 h period.There was no statistically significant difference between A-IPF values obtained at T0 and T24 (p = 0.821).However, a steady increase in %-IPF and A-IPF was seen (Figures 1 and 2), albeit it remained statistically insignificant (p > 0.05).
Furthermore, additional research to investigate the time points at which %-IPF and A-IPF exceed stability (CV% ≥ 10.0%) were carried out (n = 5) using samples with thrombocytosis (PC > 400 x 10 9 /L), thrombocytopenia (PC < 150 x 10 9 /L) as well as a PC within the normal range (150 to 400 x 10 9 /L).All samples had exceeded the set acceptance criterion of CV% < 10.0% at 72 h for both %-IPF and A-IPF.Specifically, when investigating the time point at which  stability is exceeded for each sample, %-IPF was shown to be stable in storage for 60 h and 66 h in thrombocytosis (sample 1 and 2, respectively), 66 h in thrombocytopenia (sample 3 and 4) and 18 h in a sample with a PC within normal range (sample 5).On the other hand, A-IPF was stable over 36 h in thrombocytosis (sample 1) and thrombocytopenia (sample 3), 48h in thrombocytosis (sample 2), 48 h in thrombocytopenia (sample 4) and 24 h in sample with a PC within normal range (sample 5).Overall, %-IPF showed more prolonged stability when compared with A-IPF (Supplementary Table 3).Figures 3 and 4 graphically illustrate the stability trends over 72 h for %-IPF and A-IPF, respectively, with dotted vertical lines intersecting solid lines for each sample which represent the time point where the set criterion for stability has been exceeded (i.e.CV% ≥ 10.0%).

Discussion
The quality of laboratory diagnosis includes the total testing process, i.e. reproducibility and stability which is a crucial aspect of testing to provide reliable results [17].The focus of the current study was to investigate the reproducibility and stability of %-IPF and A-IPF using the Sysmex XN-10 automated haematology analyser (Sysmex Corporation) and explore the time points at which %-IPF and A-IPF exceed stability, which, to the best of our knowledge, has not been conducted to date.
Previously, several studies have reported reproducibility and/or stability of IPF using different equipment [8][9][10][11][12]18].The earliest was the study by Watanabe et al. who had endorsed the reproducibility and stability of the IPF (both percent and absolute) using R-3000, albeit without any specific data [18].Later, using Sysmex XE-2100 (Sysmex, Kobe, Japan), Briggs et al. reported that seven samples remained stable for %-IPF over two days (0.5-48 h after sampling) at RT without any consistent increase or decrease in measurements [8].Similarly, using the same instrument, in a larger study by Ruisi et al. (n = 103), IPF was reported to be stable over 24 h period at RT [9].Unlike the current study, it is worth noting that none of these studies had reported the reproducibility and stability of IPF using the Sysmex XN-10 analyser.
A more recent study by Noel et al. had used Sysmex XN-9000 to focus on comparing IPF in EDTA and citrate samples.The group had simultaneously reported the stability of IPF over 24 h in 10 healthy and 10 routine samples to demonstrate good stability of %-IPF and A-IPF, but, unlike the current study, without the inclusion of thrombocytopenic samples, which the authors acknowledged as a limitation in their study to assessing bleeding and transfusion requirements [15].The study also did not report patient demographics (age and sex) and diagnosis, nor explored the time points at which %-IPF and A-IPF exceed stability as done in the current study.
In contrast to aforementioned studies, unreliable stability findings for %-IPF were reported over 24 h post-venesection on the same analyser, Sysmex XE-2100, by Osei-Bimpong using samples from 111 males and 89 females (age range 19-69) stored at 4 °C.This study reported a steady increase in the %-IPF (without reporting A-IPF and PC), which, after 12 h, had been significantly affected by the storage process (p < 0.0001), to an extent that could lead clinicians to misinterpret the results and thus would negate the utility of this parameter [10].Therefore, to preserve the clinical utility, Osei-Bimpong et al. later proposed an algorithm to correct for the storage effect in a study where they, once again, found a consistent linear increase (n = 29) in the %-IPF [11].
It is worth noting, the results from the two studies by Osei-Bimpong et al. [10][11], may be influenced by two factors.Firstly, a higher fluctuation of results (represented with elevated CV% values) are to be expected where %-IPF values are lower [8] (particularly when a low %-IPF accompanies severe thrombocytopenia in a sample), and their two studies reported relatively lower %-IPF without investigating repeatability and/or reproducibility of %-IPF as carried out in the present study separately to the stability over time.Secondly, these studies reported only the %-IPF, without the PC and A-IPF.When laboratory parameters are translated into clinical practice, it is preferred to use absolute (A-IPF) values (for example, in patients receiving platelet transfusions), rather than percent (%-IPF), for presenting data because %-IPF can be inordinately elevated in extremely reduced platelet counts [19].Nonetheless, all these previous studies had reported reproducibility and stability using different haematology analysers and study designs to the present study.
The results from the present study predominantly agreed with the earlier reports by Briggs et al. [8] and Ruisi et al. [9], in which using the Sysmex XE-2100, reproducibility and stability of %-IPF and A-IPF were demonstrated to be acceptable when stored at RT for at least up to 24 h.However, the current study unexpectedly reports three samples with a diagnosis of acute liver disease, undefined, and sickle cell disease that had exceeded the acceptance criterion for stability at 24 h for both %-IPF and A-IPF (Supplementary Table 2).These findings may be attributed to two possible reasons, as follows: i) either these may be due to pre-analytical or analytical errors (i.e.improper sample mixing before analysis or equipment malfunction during analysis), or ii) these measurements may be influenced by the clinical situations of these patients (acute liver disease and sickle cell disease), which requires further research.
The steady increase in %-IPF and A-IPF seen in the stability analysis of the present study (Figures 1 and 2), albeit statistically insignificant (p > 0.05) up to 24 h, is incompletely understood.The same trend is also visible in Figures 3 and 4. It is likely to be an effect of storage over time on the nucleic membrane pores of the white cells.This effect of the storage process can cause loss of cellular structural integrity, thereby, the fluorescence of the RNA released from the stored white cells can falsely increase the fluorescence signal mimicking the immature platelets, and hence %-IPF and A-IPF can be progressively overestimated in vitro over time [20].Another likely explanation could be that platelet division has been observed in vitro [21], which is suggestive of true increase of %-IPF and A-IPF during the storage process.Albeit, there were no consistent changes to the mean measurements, both increase or decrease, in the first 12 time intervals (T0 to T12) for %-IPF, and first 10 time intervals (T0 to T10) for A-IPF measurements, indicating at least 11 h and 8 h stability, respectively, without any such apparent affect.However, a small rise in the mean %-IPF and A-IPF values after T12 and T10, respectively (see Figures 1 and 2), remained statistically insignificant (p = 0.767 and p = 0.821, respectively).
Finally, the current study reports the time points at which %-IPF and A-IPF exceed stability during storage at RT in different concentrations of platelets ranging from thrombocytopenia to thrombocytosis.The study found that the time points at which %-IPF and A-IPF exceed the limit of stability range from 18 h to 66 h, with no patient sample remaining stable up to the full duration of 72 h (Figures 3 and 4 and Supplementary Table 3).

Conclusion
Based on the aforementioned conditions used in this study and the results reported, we conclude that both %-IPF and

Figure 1 .
Figure 1.Stability for %-iPf at rt (19-25 °c) using the mean values (n = 21).the time of receipt (t0) to the twelfth hour (t12) on an hourly basis, and then a final interval at 24 h (t24).time interval 1-13 in the graph represents to-t12 and then time interval 14 represents t24.

Figure 2 .
Figure 2. Stability for a-iPf at rt (19-25 °c) using the mean values (n = 21).the time of receipt (t0) to the twelfth hour (t12) on an hourly basis, and then a final interval at 24 h (t24).time interval 1-13 in the graph represents to-t12 and then time interval 14 represents t24.

Figure 3 .
Figure 3. Stability for %-iPf (n = 5) at rt (19-25 °c). the time of receipt (t0) to seventy-two hour (t72) on a six-hourly basis.time interval 1-13 in the graph represents to-t72.dotted vertical lines intersecting solid lines for each sample represent the time point at which %-iPf exceeds the set criterion for stability (cv% ≥ 10.0%).

Figure 4 .
Figure 4. Stability for a-iPf (n = 5) at rt (19-25 °c). the time of receipt (t0) to seventy-two hour (t72) on a six-hourly basis.time interval 1-13 in the graph represents to-t72.dotted vertical lines intersecting solid lines for each sample represent the time point at which a-iPf exceeds the set criterion for stability (cv% ≥ 10.0%).