Chemistry safety assessments are interpreted by using chemistry reference ranges (CRRs). Verification of CRRs is time consuming and often requires a statistical background.
We report on an easy and cost-saving method to verify CRRs.
Using a former method introduced by Sigma Diagnostics, three study sites in sub-Saharan Africa, Bondo, Kenya, and Pretoria and Bloemfontein, South Africa, verified the CRRs for hepatic and renal biochemistry assays performed during a clinical trial of HIV antiretroviral pre-exposure prophylaxis. The aspartate aminotransferase/alanine aminotransferase, creatinine and phosphorus results from 10 clinically-healthy participants at the screening visit were used. In the event the CRRs did not pass the verification, new CRRs had to be calculated based on 40 clinically-healthy participants.
Within a few weeks, the study sites accomplished verification of the CRRs without additional costs. The aspartate aminotransferase reference ranges for the Bondo, Kenya site and the alanine aminotransferase reference ranges for the Pretoria, South Africa site required adjustment. The phosphorus CRR passed verification and the creatinine CRR required adjustment at every site. The newly-established CRR intervals were narrower than the CRRs used previously at these study sites due to decreases in the upper limits of the reference ranges. As a result, more toxicities were detected.
To ensure the safety of clinical trial participants, verification of CRRs should be standard practice in clinical trials conducted in settings where the CRR has not been validated for the local population. This verification method is simple, inexpensive, and can be performed by any medical laboratory.
High prevalence rates for HIV, tuberculosis, malaria and several other infectious diseases are found in Africa.
Local clinical laboratories often rely on reference values established by the manufacturer or presented in a textbook, which are rarely specific for African populations.
In June 2009, we initiated a multi-centre phase III clinical trial to assess the safety of the antiretroviral combination tenofovir disproxil fumarate/emtricitabine (TDF/FTC) as pre-exposure prophylaxis for HIV among women in sub-Saharan Africa (FEM-PrEP trial).
The analysers, the analytical methods and the study populations differed among the study sites in the FEM-PrEP trial, making the transferability of CRRs between sites inappropriate. In European and United States-based clinical trials, it is common practice to test all study samples in a central laboratory to limit test variation across study sites.
According to the Clinical and Laboratory Standards Institute (CLSI), each laboratory should determine its own laboratory reference limits, including the CRR.
The majority of clinical laboratories collaborating in the FEM-PrEP trial had nationally- or regionally-established CRRs or used the manufacturer’s ranges prior to the study. Use of manufacturer-defined CRRs may not be appropriate for clinical trial target populations due to potentially important differences in socio-demographic characteristics, environmental context, malnutrition, dietary patterns, genetics, or infection with helminths or other parasites (such as malaria or schistosomiasis).
Our objective was to use a feasible and inexpensive method to verify previously-established CRRs and assess the impact of the final revised CRRs on the observed safety results in our phase III clinical trial.
The FHI 360 Protection of Human Subjects Committee (PHSC), institutional review boards at all study sites and applicable regulatory committees approved the study.
Women were recruited at four different sites in Bondo, Kenya; Pretoria and Bloemfontein, South Africa; and Arusha, Tanzania from June 2009 to April 2011. Because of a decision on April 18, 2011 to close the study early due to futility, verification of the CRR was not finalised for the Arusha site and is therefore not addressed in this paper. Women had to be aged between 18 and 35 years, not pregnant, and in general good health to be included in the study. The CRRs were verified at each site using a subset of 10 screened participants who were negative for: HIV, hepatitis B virus,
All laboratory activities, including specimen transport, processing, testing, result reporting and storage, were conducted in accordance with Good Clinical Laboratory Practices. At each study site, serum was collected in a plastic uncoated serum separation tube at screening, at weeks 4, 12, 24, 36, 52 and 56, and when clinically indicated. Samples were immediately taken to the on-site laboratory and were processed within two hours of collection. Quantification of AST/ALT, phosphorus and serum creatinine was performed according to the procedures described by the manufacturer and documented in site-specific standard operating procedures. The study site in Bondo performed the chemistry testing on-site using VITROS DT II (Ortho-Clinical Diagnostics, Inc., Johnson & Johnson, Buckinghamshire, United Kingdom) until May 2010 and thereafter used the VITROS 250 instrument (Ortho-Clinical Diagnostics, Inc., Johnson & Johnson, Buckinghamshire, United Kingdom). Two private laboratories performed the chemistry analysis on the samples collected at the South African study sites. Both laboratories were ISO 15189 accredited, had excellent infrastructure and had collaborated previously in multi-centre clinical trials. The study site in Pretoria shipped serum samples daily for chemistry analysis in temperature-monitored cool boxes with ice packs to the Global Clinical and Viral Laboratory (GCVL; Durban, South Africa). Upon arrival, the samples were immediately analysed using the Synchron CX5 Beckman Chemistry Analyzer (Beckman Coulter, Inc., Fullerton, California, United States). The Bloemfontein site transported the serum samples every two hours to PathCare Laboratories, located five minutes’ drive from the study site. The samples were transported at ambient temperature and were analysed immediately upon arrival using the Synchron CX5 Beckman Chemistry Analyzer. All sample transportation was validated before implementation.
After the study began, we verified the CRRs for AST, ALT, creatinine and phosphorus according to guidelines established previously by Sigma Diagnostics based on the biological variation of the analytes (no reference available). In brief, the following was performed at each study site. The ALT, AST, creatinine and phosphorus values of serum samples collected from 10 clinically-healthy participants obtained at screening were used to calculate a ‘patient mean’, after which the mean of the current reference range was determined.
Example: Manufacturer’s reference range for AST = 9 to 52 U/L:
The patient mean was compared with the established reference range mean and the percent difference between the selected samples, and the established reference mean was calculated.
Example: AST Reference Range Mean = 30.5 U/L
Selected Patient Sample Mean = 23.4 U/L
The percent (%) deviation was compared with the tolerance limit listed in the Reference Range Deviation Tolerance Limits table (
Reference range verification tolerance limits according to Sigma Diagnostics.
Analyte | % Deviation limit | Analyte | % Deviation limit |
---|---|---|---|
Alanine aminotransferase | +/− 14% | Creatinine | +/− 10% |
Albumin | +/− 8% | Glucose | +/− 8% |
Alkaline Phosphatase | +/− 20% | Total Iron | +/− 14% |
Amylase | +/− 20% | LDL / LDH | +/− 14% |
Aspartate aminotransferase | +/− 14% | LD-1 | +/− 20% |
Total Bilirubin | +/− 14% | Magnesium | +/− 16% |
Calcium | +/− 9% | Total Protein | +/− 8% |
Chloride | +/− 4% | Triglycerides | +/− 16% |
Total Cholesterol | NA | Blood Urea Nitrogen | +/− 6% |
Cholesterol, HDL | +/− 20% | Uric Acid | +/− 12% |
Creatine Kinase | +/− 20% | All Other Analytes | +/− 20% |
Creatinine Kinase-MB | +/− 20% |
HDL, high-density lipoprotein; LD-1, lactate dehydrogenase isoenzyme 1; LDH, lactate dehydrogenase; LDL, low-density lipoprotein; NA, Not Applicable.
If the percent deviation was within the listed tolerance limits, the current CRR could be used and no further action was required. Cases in which the percent deviation exceeded the tolerance limit required collection of additional values to adjust the ranges. We determined
From all sites, 20 values per analyte were available permitting us to compare the applied verification method with the CLSI guidelines.
The freeware ‘Reference Value Advisor’ (RVA) was used to perform all calculations according to the CLSI.
Initial and revised chemistry reference range values for each study site.
Site | Test | Analyser | CRR Initial |
CRR Final |
---|---|---|---|---|
Bondo | ALT | VITROS DT60/ VITROS 250 | 9–52 U/L | NA |
AST | 14–36 U/L | 2–27 U/L | ||
Creatinine | 62–106 µmol/L | 45–99 µmol/L | ||
Phosphorus | 0.81–1.45 mmol/L | NA |
||
Bloemfontein | ALT | Synchron CX5 Beckman | 10–32 U/L | NA |
AST | 10–32 U/L | NA |
||
Creatinine | 60–100 µmol/L | 38–72 µmol/L | ||
Phosphorus | 0.80–1.40 mmol/L | NA |
||
Pretoria | ALT | Synchron CX5 Beckman | 10–45 U/L | 4–26 U/L |
AST | 5–40 U/L | NA |
||
Creatinine | 60–110 µmol/L | 36–84 µmol/L | ||
Phosphorus | 0.80–1.55 mmol/L | NA |
ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRR, Chemistry reference range; NA, not applicable.
Initial refers to the CRRs that were first used by the laboratory;
Final refers to the CRR after revision;
Tests that had no revised CRRs.
Comparison of chemistry reference range from the Pretoria site obtained with the Sigma method and the Clinical and Laboratory Standards Institute method.
Site | Test | CRR Initial |
CRR Final |
CRR Final CLSI using RVA ( |
---|---|---|---|---|
Pretoria | ALT | 10–45 U/L | 4–26 U/L | 6–34 U/L |
AST | 5–40 U/L | NA |
13–29 U/L | |
Creatinine | 60–110 µmol/L | 36–84 µmol/L | 44–88 µmol/L | |
Phosphorus | 0.80–1.55 mmol/L | NA | 0.84–1.53 mmol/L |
ALT, alanine aminotransferase; AST, aspartate aminotransferase; CLSI, Clinical and Laboratory Standards Institute; CRR, Chemistry reference range; NA, not applicable; RVA, Reference Value Advisor freeware.
Initial refers to the CRRs that were first used by the laboratory;
Final refers to the CRR after revision;
Tests that had no revised CRRs.
The grading of adverse events based on laboratory abnormalities was performed in accordance with the DAIDS grading table and therefore relied on the ULN. The final revised CRRs were applied from the time of establishment; previous adverse events were not re-graded during the study.
Toxicity grades for AST/ALT and creatinine using initial versus final reference ranges.
Toxicity | Grades | Final CRR |
|||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Grade 0 |
Grade 1 |
Grade 2 |
Grade 3 |
Grade 4 |
Total |
||||||||
Tests | % | Tests | % | Tests | % | Tests | % | Tests | % | Tests | % | ||
Initial CRR | Grade 0 | 8170 | 95.1 | 206 | 2.4 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 8376 | 97.5 |
Grade 1 | 0 | 0.0 | 144 | 1.7 | 0 | 0.0 | 0 | 0.0 | 169 | 2.0 | |||
Grade 2 | 0 | 0.0 | 0 | 0.0 | 27 | 0.3 | 0 | 0.0 | 40 | 0.5 | |||
Grade 3 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 3 | 0.0 | 0 | 0.0 | 3 | 0.0 | |
Grade 4 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 3 | 0.0 | 3 | 0.0 | |
Total | 8170 | 95,1 | 350 | 4.1 | 52 | 0.6 | 16 | 0.2 | 3 | 0.0 | 8591 | 100.0 | |
Initial CRR | Grade 0 | 8128 | 92.7 | 412 | 4.7 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 8540 | 97.4 |
Grade 1 | 0 | 0.0 | 186 | 2.1 | 0 | 0.0 | 0 | 0.0 | 205 | 2.3% | |||
Grade 2 | 0 | 0.0 | 0 | 0.0 | 16 | 0.2 | 0 | 0.0 | 18 | 0.2 | |||
Grade 3 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 1 | 0.0 | 0 | 0.0 | 1 | 0.0 | |
Grade 4 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 2 | 0.0 | 2 | 0.0 | |
Total | 8128 | 92.7 | 598 | 6.8 | 35 | 0.4 | 3 | 0.0 | 2 | 0.0 | 8766 | 100.0 | |
Initial CRR | Grade 0 | 8458 | 97.4 | 43 | 0.5 | 0 | 0.0 | 0 | 0.0 | 8502 % | (97.9 | ||
Grade 1 | 0 | 0.0 | 167 | 1.9 | 0 | 0.0 | 0 | 0.0 | 180 | 2.1 | |||
Grade 2 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | |
Grade 3 | 0 | 0.0 | 0 | 0.0 | 0 | 0.0 | 1 | 0.0 | 0 | 0.0 | 1 | 0.0 | |
Grade 4 | NA | NA | NA | NA | 0 | 0.0 | NA | ||||||
Total | 8458 | 97.4 | 210 | 2.4 | 14 | 0.2 | 1 | 0.0 | 0 | 0.0 | 8683 | 100.0 |
Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRR, Chemistry reference range; NA, not applicable.
†, Final refers to the CRR after revision; ‡, Initial refers to the CRRs that were first used by the laboratory; §, Tests that had no revised CRRs.
The laboratory abnormality frequency for ALT, AST, creatinine, and phosphorus between the two study groups (placebo vs. TDF/FTC) using the initial and final CRRs were compared and are presented in
Laboratory abnormality frequency tables based on the initial versus final chemistry reference ranges.
Parameter | Abnormality | Placebo ( |
TDF/FTC ( |
Total initial/final CRRs |
|
---|---|---|---|---|---|
Number of events | Number of events | Number of events | Total number and percentage missed using initial ranges | ||
ALT | Grade 1 | 57/84 | 60/131 | 117/215 | 98 = 45.6% |
Grade 2 | 7/8 | 14/16 | 21/24 | 3 = 12.5% | |
Grade 3 | 2/6 | 2/4 | 4/10 | 6 = 60.0% | |
Grade 4 | 2/2 | 1/2 | 3/4 | 1 = 25.0% | |
4/8 | 3/6 | 7/14 | 7 = 50.0% | ||
AST | Grade 1 | 79/174 | 81/205 | 160/379 | 219 = 57.8% |
Grade 2 | 7/13 | 10/21 | 17/34 | 17 = 50.0% | |
Grade 3 | 1/1 | 0/2 | 1/3 | 2 = 66.7% | |
Grade 4 | 0/0 | 1/1 | 1/1 | 0 = 0.0% | |
1/1 | 1/3 | 2/4 | 2 = 50.0% | ||
Creatinine | Grade 1 | 54/67 | 81/85 | 135/152 | 17 = 11.2% |
Grade 2 | 0/2 | 0/3 | 0/5 | 5 = 100.0% | |
Grade 3 | 0/0 | 1/1 | 1/1 | 0 = 0.0% | |
Grade 4 | 0/0 | 0/0 | 0/0 | 0 = 0.0% | |
0/2 | 1/4 | 1/6 | 5 = 83.3% | ||
Phosphorus | Grade 2 | 215/215 | 203/203 | 418/418 | 0 = 0.0% |
Grade 3 | 43/43 | 50/50 | 93/93 | 0 = 0.0% | |
Grade 4 | 0/0 | 0/0 | 0/0 | 0 = 0.0% | |
43/43 | 50/50 | 93/93 | 0 = 0.0% |
ALT, alanine aminotransferase; AST, aspartate aminotransferase; CRR, Chemistry reference range; TDF/FTC, tenofovir disproxil fumarate/emtricitabine.
Initial refers to the CRRs that were first used by the laboratory. Final refers to the CRR after revision.
There were no significant differences in the proportion of women experiencing toxicities in the TDF/FTC and placebo groups based on the initial CRR (results not shown). Using the revised CRR, however, a significantly higher percentage of women in the TDF/FTC group experienced ALT grade 1 or higher toxicities (
The CLSI guidelines recommend the establishment of CRRs with at least 120 reference individuals using a non-parametric ranking method or, as an alternative, a robust method with a minimum of 20 samples from qualified reference individuals when there are sample size constraints. It was not feasible in the FEM-PrEP trial to recruit reference individuals prior to the initiation of the trial, as the study sites were research centres that did not see routine patients. Therefore, we verified the existing CRRs using specimens collected at screening using the Sigma verification procedure. This method is simple, does not require statistical expertise, is less time-consuming, inexpensive and can easily be implemented by any laboratory. We also examined the impact of the new reference ranges on the toxicity grading. When looking at our data, a total of 9 ALT/AST grade 3 or higher, 5 serum creatinine grade 2 toxicities and many grade 1 toxicities would have been missed if the original CRRs were used.
Laboratories are essential for both the detection and prevention of diseases. In clinical trials, laboratories play a crucial role in endpoint measurement. In the FEM-PrEP trial, the laboratory safety endpoints were based on chemistry parameters to detect liver and kidney toxicities. There were no additional clinical or laboratory costs involved in the verification process since the chemistry tests were a required screening procedure. The major disadvantage of using specimens collected at screening was that study participants could be excluded or included erroneously through misclassification of toxicity grades during the time of CRR verification and adjustment. However, in FEM-PrEP there were no instances of discordant eligibility classification when applying the pre-existing versus final verified CRRs due to predefined inclusion criteria, which required that creatinine be < 1.5 mg/dL and hepatic function tests be < 2x ULN.
Immunohaematological reference ranges are now well defined in Asia and Africa, and different studies have reported the need for population-specific clinical chemistry reference ranges.
Comparison of chemistry reference ranges from the three study sites to those reported in previous publications.
Analyte | Unit | Bondo (Kenya) | Pretoria (South Africa) | Bloemfontein (South Africa) | Uganda |
Rwanda, Uganda, Kenya and Zambia |
Kenya |
Ghana |
Kenya |
Tanzania |
|
---|---|---|---|---|---|---|---|---|---|---|---|
U/L | LLN | 9 | 4 | 10 | 5.3 | 8 | 10.7 | 6 | 8.6 | 0 | |
ULN | 52 | 26 | 32 | 39.9 | 61 | 61.3 | 51 | 47 | 44.9 | ||
U/L | LLN | 2 | 5 | 10 | 11.4 | 14 | 13.5 | 13 | 13.1 | 13.5 | |
ULN | 27 | 40 | 32 | 28.8 | 60 | 48.5 | 48 | 38.1 | 35.2 | ||
µmol/l | LLN | 45 | 36 | 38 | 44.2 | 47 | 52.4 | 47 | 51 | 40 | |
ULN | 99 | 84 | 72 | 79.6 | 109 | 96.8 | 110 | 91 | 81 | ||
mmol/L | LLN | 0.81 | 0.80 | 0.80 | 0.81 | ND | ND | 0.8 | ND | ND | |
ULN | 1.45 | 1.55 | 1.40 | 1.81 | ND | ND | 1.5 | ND | ND | ||
♀ (18–35y) | ♀ (18–35y) | ♀ (18–35y) | ♀ (18–56y) | ♂ + ♀ (18–34y) | ♀ (18–34y) | ♀ (18–59y) | ♀ (18–55y) | ♀ (19–48y) | |||
VITROS 60/250 | Synchron CX5 Beckman Chemistry Analyser | Synchron CX5 Beckman Chemistry Analyser | Roche Cobas Integra 400 plus Analyser | Vitalab Selectra E Clinical Chemistry Analyser | Roche Cobas Integra 400 plus Analyser | Vitalab Selectra E Clinical Chemistry Analyser | Roche Cobas Integra 400 plus Analyser | Roche Cobas Integra 400 plus Analyser | |||
Sigma | Sigma | Sigma | Wilcoxon | CLSI | Statistics | CLSI | CLSI | CLSI | |||
2014 | 2014 | 2014 | 2008 | 2009 | 2011 | 2012 | 2008 | 2008 |
CLSI, Clinical and Laboratory Standards Institute; LLN, lower limit of normal; ND, Not Done; ULN, upper limit of normal.
Our study has several limitations. Ideally, verification of reference ranges should be conducted before trial initiation. It is also possible that the number of specimens (10) required for initial verification by this method were too few. For example, CLSI recommends a set of 20 reference specimens and replacement of outliers if necessary. We compared the applied method with the CLSI guidelines and obtained similar results, except for one parameter (ALT) in one study site (Pretoria) which should not have been corrected according to CLSI. As we had more than 120 reference values available (
We detected a large number of toxicities that would not have been identified using the pre-existing CRRs due to the decrease in the ULN for hepatic and renal parameters. Overall, we developed more population-appropriate CRRs that may have improved the clinical safety management of study participants. In conclusion, establishing local reference ranges is necessary to comply with the high-quality standards of Good Clinical and Laboratory Practices. Unfortunately, not all laboratories have the resources necessary to establish local reference ranges; therefore, verification of existing reference ranges offers a good alternative. Methods such as the former Sigma method or freeware including robust, transformation and non-parametrical methods can be applied on reference samples sets without additional costs and in the absence of sophisticated statistics by any laboratory performing chemical analysis.
During several laboratory supervision visits, the correctness of the values that were used to verify the CRR was checked with the raw data. All laboratories worked according to good clinical laboratory practice guidelines and two laboratories were also ISO 15189 accredited.
We report here on an easy method that uses 10 values to verify the Reference Ranges and that can be implemented in any laboratory without need for statistical expertise. We also show that determining the CRR before the start of a clinical trial is imperative to ensure that all toxicities found in the study are graded correctly. With this study, we also compared the Sigma method with the CLSI guidelines and obtained similar results. Our reported CRR are also within range with what has been previously found.
This study was conducted in the context of the FEM-PrEP clinical trial. We thank all the women who participated in the clinical trial and all study staff who worked on the study.
The authors declare that they have no financial or personal relationship(s) that may have inappropriately influenced them in writing this article.
The main study was made possible through grants funded by the United States Agency for International Development (USAID), the Contraceptive and Reproductive Health Technologies and Research Utilization Program, and the Preventive Technologies Agreement No GHOA000900016-00. Early support was also provided by the Bill and Melinda Gates Foundation. Gilead Sciences, Inc. donated the TDF-FTF and placebo. Views expressed in this publication do not necessarily reflect those of FHI 360 or the agencies funding the study.
I.D.B. and T.C. wrote the first draft of the manuscript. D.T., J.M., K.N., J.D. and L.V.D. revised and edited the text. T.C. and C.V.C. created the experimental design, and D.T. performed the statistical analysis. W.A., L.M., and E.-M.B. generated the data. All authors revised and approved the present version of the manuscript.