Effect of Temperature and Time on Fecal Hemoglobin Stability in 5 Fecal Immunochemical Test Methods and One Guaiac Method Open Access

The following terminology is used throughout. Kit refers to the package of components, including collection devices, required for the collection and analysis of Hb in fecal samples. Collection device is the component of a kit to which the feces are added and stored until analysis. Spiked feces is a fecal sample to which Hb has been added in the laboratory. Inoculated device is a collection device to which feces (spiked or nonspiked) has been added for analysis of Hb.

Three in vitro studies were conducted from February to April 2012 to assess Hb stability. Study 1 was a controlled study of inoculated devices stored at various controlled temperatures (frozen, refrigerated, ambient, and elevated) and over time up to 60 days. Study 2 was a controlled study of inoculated devices exposed to several freeze-thaw cycles. Study 3 was a field study in which inoculated devices were transported by courier or postal services under uncontrolled temperature conditions, which was intended to simulate conditions that inoculated devices would experience in the CRC screening program in Ontario, Canada.

This study was reviewed and approved by the Sunnybrook Research Institute (Toronto, Ontario, Canada) research ethics board.

Table 1 summarizes the characteristics of the kits included in the current study. The FIT kits were selected for inclusion if they were approved for use by Health Canada and if they were suitable for use in an organized, population-based CRC program (ie, they were not point-of-care kits). The following FIT kits were included: (1) Hb NS-Plus, manufactured by Alfresa Pharma Corporation (Chuo-ku, Osaka, Japan); (2) OC-Sensor Diana, manufactured by Eiken Chemical Co, Ltd (Taito-ku, Tokyo, Japan); (3) Hema-Screen SPECIFIC, manufactured by Immunostics, Inc (Eatontown, New Jersey); (4) FOB Advanced, manufactured by ulti med (Ahrensburg, Germany); and (5) Hemoccult ICT, manufactured by Beckman Coulter, Inc (Brea, California). The gFOBT (Hema-Screen Guaiac, Immunostics) currently used in the Ontario, Canada, CRC screening program was also included.

For all studies, fresh human feces, without preservatives and from multiple donors, were used. Feces were spiked with human Hb from ethylenediamine tetra-acetic acid whole blood specimens. Initial Hb concentrations were determined with the Sysmex XE-2100 series automated hematology analyzers (Sysmex Corporation, Kobe, Hyogo Prefecture, Japan) using the sodium lauryl sulfate method, calibrated to International Council for Standardization in Hematology/World Health Organization international hemoglobincyanide standard (70600). Hemolysates were prepared by diluting whole blood 10-fold with deionized water. The Hb was transferred to the fecal material, and a homogeneous mixture was obtained by stirring the feces with wooden sticks. Collection devices were inoculated with feces at room temperature within 30 minutes of spiking with Hb. Using the procedure described, acceptable and reproducible recoveries of Hb were obtained. When measured with the OC Diana kit, the range of Hb recovery was 85.4% to 112.7% in 16 replicates. Mean (SD) recovery was 103.4% (9.7%).

To date, there is no universal or standardized approach to assess Hb stability.^19,20  Although some manufacturers do provide estimates of Hb stability in their product inserts, those estimates cannot be reliably compared across brands because the methods to determine stability were not described in the package inserts and time frames varied. For the purposes of our analysis, we measured Hb stability using separate, but standardized, approaches for the comparison of quantitative kits alone and for comparison of all kits included in the study. These 2 approaches are described below.

For comparison of Hb stability in the collection devices of the quantitative kits only, the duration of stability in study 1 was defined as the time until the percentage of Hb recovery was reduced (from 100%) by less than 2 times the between-run precision, as determined from internal studies. The between-run precision was approximately 13% and 9% at concentrations of 33 and 100 μg Hb/g feces, respectively. Therefore, mean recovery of 80% or more was used as a cutoff for stability when comparing the 2 quantitative kits.

Because qualitative FIT and gFOBT test results are read in a binary fashion, the above approach could not be used. To compare Hb stability across all kits (both qualitative and quantitative), we defined the duration of stability as the time that elapsed until the probability of the result remaining positive dropped below a prespecified threshold. For this analysis, quantitative kit results were categorized as having positive or negative results relative to the manufacturer's suggested positivity threshold (20 μg Hb/g feces for both kits). Stability at 3 prespecified thresholds (80% probability of remaining positive, 70%, and 50%) was measured. The clinical implications of these thresholds are not known. However, the only other study¹⁷  to establish a threshold for stability used 50%, so our thresholds match or exceed those used by others.

For all 3 studies, results were defined as positive if the Hb concentration was above the manufacturer's recommended positivity threshold (see Table 1) for quantitative kits or if the visual band was detected for qualitative kits.

Preliminary in-house studies showed that the analytic sensitivities of the kits were 20 μg Hb/g of stool (NS Plus and OC Diana), 50 μg Hb/g of stool (Immunostics FIT), 100 μg Hb/g of stool (ulti med), 500 to 1000 μg Hb/g of stool (Beckman ICT), and 500 μg Hb/g of stool (Immunostics Guaiac). Feces were spiked to produce Hb concentrations from 25 to 3000 μg Hb/g of feces. As the analytic sensitivities (which reflect the lower limit of Hb detection) varied across kits, concentration ranges appropriate to the analytic sensitivity of each kit were selected.¹⁸  Final Hb concentrations of 25, 50, 100, and 200 μg Hb/g of feces were used for the NS Plus, OC Diana, and Immunostics FIT kits; 100, 200, 500, 1125, and 1500 μg Hb/g of feces were used for the ulti med kit; and 500, 1125, and 1500 μg Hb/g of feces were used for the Beckman ICT and Immunostics Guaiac kits.

Multiple inoculated devices were prepared for each kit (33 sets per kit) using feces spiked with each of the Hb concentrations specified above for that particular kit. The first set was analyzed immediately at day 0 (baseline). The other 32 sets were stored under the following temperature conditions: frozen, −20° to −15°C (−4° to 5°F) (8 sets); refrigerated, 2°C to 8°C (36°F to 46°F) (8 sets); ambient, 20°C to 22°C (68°F to 72°F) (8 sets); and elevated, 45°C (113°F) (8 sets). For each temperature condition, a set was removed and analyzed at 4, 7, 10, 14, 21, 28, 45, and 60 days after baseline. Each inoculated device was tested only once during this study.

We calculated the median percentage of Hb recovery at each time point in each of the temperature conditions for each of the 2 quantitative kits. We then examined Hb stability in 2 ways: comparison of the inoculated devices from the quantitative kits only, and comparison of the inoculated devices from all kits, both quantitative and qualitative.

For the comparison of Hb stability in collection devices of the quantitative kits only, the percentage of Hb recovery compared with the baseline were first calculated for each temperature condition and at each time point. As noted above, recovery of 80% or more was used as a cutoff for stability. Under each of the 4 temperature conditions, the relationship between the percentage of Hb recovery and time was first modeled using a linear regression model. Quadratic and cubic terms were included in the regression if they improved the model fit based on likelihood ratio tests.²¹  A generalized estimating equation approach was used to account for any repeated measurements taken on the same sample.²²  Once the estimates of the regression coefficients were obtained, those values were then used to predict the time in days and the 95% CI that Hb recovery would drop below 80% compared with the baseline.

For the comparison of Hb stability in the collection devices across all kits, results for both the quantitative and qualitative kits were expressed as binary results (negative/positive) only. As noted above, stability was defined as the time required for the probability of results remaining positive to drop below 3 prespecified thresholds. Each threshold was modeled separately, starting with the 80% threshold. Under each of the 4 temperature conditions, the relationship between a positive result (yes/no) and time was modeled using a logistic regression model. As above, quadratic and cubic terms were included where appropriate, and a generalized estimating equation approach was used. Once the estimates of the regression coefficients were calculated, those values were used to predict time in days and the 95% CI when the probability of results remaining positive dropped below 80%. These analyses were repeated for the other 2 probability thresholds, specifically, the time at which the probability of results remaining positive dropped below 70% and below 50%.

In real-world screening programs, patient samples may be exposed to variable temperatures and conditions, including freezing and thawing, during the transport of the samples from the patient to the testing laboratory. This study was designed to determine the effect of freezing and thawing on samples in the collection devices. The stability of Hb during freezing and thawing was studied in the NS Plus, OC Diana, and ulti med collection devices. (Kits with single-use collection devices could not be used for this study.) Fecal samples were spiked to produce samples with Hb concentrations ranging from 25 to 1200 μg Hb/g of feces, in keeping with the ranges for these 3 kits used in study 1. Inoculated devices were prepared with samples at Hb concentrations appropriate to the analytic sensitivity of each type of kit, as described in study 1. Inoculated devices (NS Plus, n = 12; OC Diana, n = 12; ulti med, n = 12) were analyzed immediately (baseline), and the same devices were then analyzed after each of 3 freeze-thaw cycles. Frozen storage conditions were −20°C to −15°C (−4°F to 5°F). Samples were thawed and refrozen after analysis at 4, 7, and 10 days after the initial storage in the freezer. Inoculated devices were kept frozen for 4, 3, and 3 days between thawing, testing, and refreezing for cycles 1, 2, and 3, respectively. The number of cycles and duration were selected because they were likely to reflect anticipated real-world conditions.

For the analysis, the median percentage of Hb recovered compared with baseline values was calculated for NS Plus and OC Diana after each of the freeze-thaw cycles. For ulti med, the percentage of inoculated devices remaining positive after each cycle was calculated.

A total of 30 fecal samples were freshly spiked with Hb at the testing location to achieve final Hb concentrations of 0 μg Hb/g of feces (n = 6), 30 μg Hb/g of feces (n = 3), 60 μg Hb/g of feces (n = 6), 125 μg Hb/g of feces (n = 3), 500 μg Hb/g of feces (n = 6), and 2000 μg Hb/g of feces (n = 6). Two batches of inoculated devices using those samples were prepared for each of the 6 kits. One batch of inoculated devices (set A, 30/kit for 180 inoculated devices) was analyzed at the testing location to obtain baseline values. The second batch of inoculated devices (set B, n = 180) was packaged for transportation to 3 distribution centers (Thunder Bay, Ottawa, and London; all in Ontario, Canada). The packages were prepared without any temperature control, that is, no cooling packs, but simply with cushioning material to prevent excessive contact within the package. The study was arbitrarily designed to have 10 packages for each center. Each package contained 6 inoculated devices (1 for each of the 6 kits), for a total of 60 devices. Each center received the same distribution of Hb concentrations.

For each center, thermometers were added to 9 of the 10 packages to determine minimum and maximum temperatures during transport and a continuous temperature data logger was placed in the 10th package. The batches of packages were delivered from the testing location to the 3 distribution centers by courier. From each of the distribution centers, the 10 packages were individually mailed back to the testing location, via Canada Post regular mail.

Upon receipt at the testing location, the temperature monitors were retrieved, the transportation temperatures were recorded, and the inoculated devices were analyzed. This study was performed in February and March 2012.

Two analyses were performed. In the first, the percentage of kits in which the results of the before-transportation and after-transportation tests (set B) agreed with the expected result (based on the amount of Hb spiked into each of the initial 30 fecal samples, set A) was calculated. In the second, the percentage of kits in which the result after transportation agreed with the result from before transportation (same kits, from set B) was calculated.

The median percentage of Hb recoveries in the collection devices of the 2 quantitative kits over time and at varying temperatures are shown in the Figure, a and b. Inoculated devices stored under refrigerated conditions had the greatest sample stability for both NS Plus and OC Diana. Freezing the devices produced an initial decrease of 20% in Hb recovery, but then Hb recovery remained stable over time thereafter for that temperature condition. There were decreasing recoveries with increasing temperature and increasing time for the other 2 temperature conditions. Across all temperature conditions, there was a tendency toward better stability for NS Plus.

Results from the modeling comparing the 2 quantitative kits are shown in Table 2. Hemoglobin stability was similar, although there was a tendency toward longer stability for NS Plus at most temperatures. At ambient temperatures for both devices, the predicted stability, using time to less than 80% Hb recovery as the standard, was less than a week.

Results comparing all 6 kits are shown in Table 3. The Hb stability in the 2 quantitative kits—NS Plus and OC Diana—was poor compared with the devices of the qualitative FIT and guaiac kits in the frozen condition when an 80% threshold was used to define stability. The observed differences between quantitative and qualitative kits were reduced for the 70% and 50% thresholds. Again, most kits performed well under refrigerated conditions, regardless of the threshold used, with the exception of Beckman ICT and ulti med kits.

Under ambient conditions, Immunostics FITs tended to have the strongest performance, although the quantitative kits also performed reasonably well, particularly at the 70% and 50% thresholds. The Immunostics Guaiac also performed well, whereas the ulti med and Beckman ICT kits performed quite poorly at ambient temperature.

None of the kit collection devices performed well at high (45°C) temperatures, with the exception of the Immunostics Guaiac (19.8 days at 80% threshold).

As illustrated in Table 4, of the 3 kits tested, the NS Plus kit provided the most consistent and acceptable recovery after up to 3 freeze-thaw cycles. Recovery with the OC Diana kit declined steadily through the 3 freeze-thaw exposures. For the ulti med kit, the only qualitative test that could be included in this study, significant reduction in recovery was observed even after one cycle.

During transportation from the 3 distribution centers, inoculated devices were exposed to average temperatures of 18.4°C, 18.8°C, and 17.0°C. The lowest and highest temperatures encountered were −14°C and +26°C, respectively. The number of freeze-thaw cycles encountered during transport varied depending on the distribution center to which the devices were sent (one center had none, one center had 3, and one center had 1). Total transportation times varied from 5 to 12 days.

Table 5 summarizes the 2 analyses of agreement in the field study. Both NS Plus and OC Diana exhibited good agreement before and after transportation and with the expected result. Immunostics FIT and ulti med had good agreement before transportation, in keeping with their known analytic sensitivities, which should be able to detect the lower Hb concentrations in the inoculated devices. However, the agreement for both was poor between the before and after transportation tests, indicating degradation of Hb (ulti med more than Immunostics FIT). The Beckman ICT and Immunostics Guaiac had poor agreement before transportation, in keeping with their known poorer analytic sensitivities. However, agreement was good between the before and after transportation tests, suggesting that Hb remained stable in the collection devices of those kits during transportation.

Using a standardized and rigorous methodology, we found important variations in Hb stability across 5 commercially available FIT kits suitable for use in organized CRC screening programs and 1 gFOBT kit currently in use in the Ontario, Canada, program. The Immunostics FIT appeared to be the best-performing qualitative kit, whereas, among the quantitative kits, NS Plus tended to perform slightly better than OC Diana. As temperatures increased above freezing, Hb recoveries decreased over time for all kits. Freezing samples resulted in a small initial loss of recovery in the collection devices of the quantitative kits. However, Hb concentration was then maintained across all time points. Some loss of recovery during the transport process was observed in the field study. The current study did not find uniformly better stability in FIT kits using dry collection devices (Immunostics FIT and Beckman ICT) compared with wet collection devices.

Although the lack of Hb stability over time, particularly at higher temperatures, is well documented,^{15,16,18,23,24}  and seasonal variation in positivity rates has been previously reported,^13,25  there are only a few reports comparing the Hb stability of different FIT kits.^17,18  Interestingly, one of those reports¹⁷  also compared NS Plus and an OC product and, similar to our results, found that the NS Plus tended to exhibit slightly better stability. To our knowledge, Immunostics FIT has not been compared with other FITs for stability. In the current study, we report a decrease in Hb concentration of about 1.4%/d and about 2.1%/d at ambient temperature (20°C–22°C) for the NS Plus and the OC Diana, respectively. One other study¹⁸  has also reported Hb instability for FITs at ambient temperatures, ranging from 1.7%/d to 7.8%/d, depending on the kit. Other studies have shown important differences in other performance characteristics for CRC screening across FIT kits.^26–29  In sum, these studies highlight a key message: not all FIT kits are the same; as such, selection of FITs for use in organized screenings should include carefully consideration of data on performance and quality control.

There are few data on the effect of freezing on Hb stability in FIT collection devices, although a study from the Netherlands reported reduced FIT detection rates in the winter.²⁵  We noted a drop in Hb concentration in the quantitative collection devices with freezing, but, once frozen, the recovery for the quantitative kits was relatively constant for up to 60 days. However, the effect of sequential freezing and thawing varied across the 3 kits we evaluated, with the NS Plus performing best. Our findings suggest that the process of freezing and thawing (which may occur during kit transport to the laboratory), and not necessarily the time frozen, may decrease Hb recovery.

Our field study was conducted outside a controlled laboratory setting, and as a result, there was considerable variation in important parameters, such as temperature, transportation time, freezing, and thawing. Although there may be limitations to this “real-world” approach in generalizability, important lessons about Hb stability have been learned from other similar studies.^13,25  Although there was some loss of recovery during the transport process in this field study, in spite of long transit times of up to 12 days and variations in temperature conditions (−14°C to +26°C), the 2 quantitative kits performed better (Hb remained stable in their collection devices across the full range of Hb concentrations tested) than the other kits. A number of studies have examined the potential clinical significance of exposure to higher temperatures¹³  and delayed sample returns.^15,16  Unfortunately, we were not able to perform the field study during the summer when temperatures are expected to be hotter. Nevertheless, data from the current study suggest that the collection devices of the quantitative kits and some devices of the qualitative kits appear reasonably robust and may be suitable for postal transport in climates similar to that in Canada, at least in winter.

In wet collection devices, such as those used by some FIT kits, the rate of Hb degradation has been reported to be greater than that in dry collection devices, perhaps because of bacterial degradation of Hb in moist environments.^11,14  The current study shows Hb stability in the collection devices of the Immunostics FIT and Immunostics Guaiac kits, which use a dry collection device, was, in general, longer than that of the other FIT kits. By contrast, Beckman ICT, which also uses a dry collection device, underperformed relative to some of the other kits using wet collection devices. These findings suggest that claims of superior sample storage capability may not be generalizable across all dry collection devices.

There are strengths and weaknesses to the current study. Because few data comparing Hb stability in collection devices across FIT kits are available, we attempted to address that issue with a novel, transparent, and easily reproducible methodology. To compare across kits, we established a definition of stability: for quantitative kits, time until Hb concentration decreased to less than 80%, a value selected based on internal studies of between-run precision; and for qualitative kits, time until the probability of remaining positive dropped below 3 prespecified thresholds. This approach is similar to that used by another study from the United Kingdom,¹⁷  although that study used a larger drop in Hb (≥50% from baseline) to define a loss of stability. A second study from France¹⁸  simply reported the percentage of change in Hb concentration by day because the authors felt they could not use regression techniques because of differences among kits in the distribution of the data. The clinical significance of different thresholds is not known and likely varies with the Hb concentration of the fecal sample and the analytic sensitivity of the particular FIT, but use of a threshold allows a common reference with which to compare kits. Unlike the 2 other studies mentioned, we used regression methods appropriate to the distribution of the data, allowing us to obtain robust summary measures of stability over time; in addition, the use of generalized estimating equations accounted for multiple measures on the same kit.

One of the challenges we faced in developing this methodology was that the kits had significantly different analytic sensitivities. We addressed that issue by selecting fecal Hb concentrations appropriate to the reported analytic sensitivity of each kit. It was observed that some of the unspiked samples gave unexpected positive results with the Immunostics and ulti med kits, which may indicate false-positive results in some of those samples. A second limitation of this study was that the data for the samples in these kits may reflect the stability of an interferent rather than that of Hb. A third limitation was that in August 2012, the OC Diana buffer was updated; therefore, our results may not be generalizable to the kit currently in use. A fourth potential limitation was that the design of the study offered no opportunity to correlate decreases in Hb concentrations with changes in polyp or cancer detection rates; therefore, we cannot comment on the clinical impact of the observed declines in Hb stability from the different kits. Data from recent studies have shown that there can also be important differences in clinical outcomes across FIT kits, even when using the same positivity threshold.²⁶ 

The use of a standardized methodology to compare Hb stability is a strength of this study. Although the clinical impact of the differences we report are not known, it may be reasonable to assume that kits that performed relatively poorly in our study may be likely to also do so in the real world. Our approach confers a benefit when comparing products for Hb stability because the reported stabilities in manufacturers' inserts almost certainly use various methodologies. In so doing, our study addresses a recent call for a standardized methodology to measure Hb stability in FIT collection devices.¹⁹ 

In summary, the current study reports considerable variation in the performance of FITs for Hb stability. We found that the Immunostics FIT was the best performer among the qualitative kits. Both quantitative kits performed well, although the NS Plus appeared to have a slight edge over the OC Diana in our study. It is not known whether that difference will be maintained now that OC Diana has been updated with a new buffer. Because not all FIT methods are the same, the current study provides important performance data on the FIT methods evaluated that will assist CRC screening programs and providers in FIT selection. Finally, we describe a novel analytic approach to the comparison of Hb stability across kits, which can be used to address the well-recognized need for standardization in this area.