CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL cholesterol, and LDL cholesterol. All of these protocols state: The CRMLN strongly recommends that manufacturers set asde and store (at -70 C or lower) addtonal alquots of each fresh sample (volume consstent wth analytcal system requrements). These samples can be used for reanalyss f changes n calbraton are requred to meet certfcaton crtera. When new lots of calbrators, materals, or reagents are prepared, these frozen samples can provde an mportant lnk to the accuracy base durng overlap analyses f a frozen versus fresh comparson has been performed. Ths protocol s desgned as a gudelne for manufacturers to use n comparng results from frozen samples versus fresh samples. The CRMLN wll not evaluate the data collected from sample stablty studes. Ths protocol s provded as a gudelne for manufacturers who would lke to save addtonal alquots of samples used n CRMLN certfcaton protocols. It s n the best nterest of the users of ths protocol to follow t carefully so that undue errors are not generated n future certfcaton attempts. Qualty Control The user of ths protocol wll need to know the coeffcent of varaton (CV) of the analytcal method to be able to determne a sample sze for the study. Ths nformaton s obtaned from the qualty control (QC) data. In addton to provdng a CV for determnng the sample sze for the study, the QC s the key to havng a vald stablty study. The manufacturer must have a stable QC system n place before begnnng the stablty study. A tact assumpton n QC s that the materal s stable over tme. The best materal for QC wll be one that has been shown to be stable for the duraton of the study. Ths establshes the method s stablty base. The materal s also expected to be stable over the tme frame beng studed so that the method s stablty can be verfed. Before begnnng, the manufacturer should have collected QC data that covers the same length of tme planned for the study. It s advsable that the QC materal be measured n addtonal runs durng the course of tme that the test samples are frozen to nsure that the method remans stable durng the study. Ths wll gve the study the statstcal power to demonstrate that any changes observed are due to changes n the fresh-frozen materals and not due to changes n the analytcal method. It s also advsable to have data from several levels of QC materals to determne f the varance s unform over the analytcal range beng studed. CRMLN Sample Stablty Protocol October 2004 Page 1
The QC characterzaton data should, deally, be collected usng a sngle lot of calbrator and a sngle lot of reagent. Mnmally, ths same lot of calbrator must also be used for the stablty study. Statstcal Approach The number of samples requred to adequately detect a dfference depends on the CV of the analytcal method. For 80% power to detect a 1% dfference, the followng numbers of samples are needed for varous analytcal CVs: CV # samples CV # samples CV # samples 1.0 18 2.0 65 3.0 144 1.1 22 2.1 72 3.1 153 1.2 25 2.2 78 3.2 163 1.3 29 2.3 86 3.3 173 1.4 33 2.4 93 3.4 184 1.5 38 2.5 101 3.5 195 1.6 43 2.6 109 3.6 206 1.7 48 2.7 117 3.7 217 1.8 53 2.8 126 3.8 229 1.9 59 2.9 134 3.9 241 The t-test wll be used to analyze the data. The t-test uses an estmate of varance (standard devaton, SD) n the calculaton, not CV. Use of the t-test assumes that the varance s unform across the analytcal range. If the varance s not unform across the analytcal range (.e. f the SD has concentraton dependence), then the range must be dvded nto concentraton regons wth unform varance. Each concentraton regon wll requre the approprate number of samples from the table above, dependng on the CV of the ndvdual regon. Ths protocol s wrtten for a method wth unform varance and an analytcal CV of 1%. If the method has a larger CV, then both the number of samples per day and the number of days when samples are collected wll necessarly need to be ncreased. Protocol The protocol s wrtten for serum. However, t can be easly adapted for plasma f that s the matrx used wth the analytcal system beng evaluated. Three tme frames are ncluded n the protocol frozen for 7, 30, and 60 days. However, the protocol can also be adapted to nclude CRMLN Sample Stablty Protocol October 2004 Page 2
addtonal or dfferent ntervals. If addtonal duratons are to be evaluated, a larger volume of each ndvdual sample wll be needed. Conversely, f fewer tme frames are to be evaluated, a smaller volume of each ndvdual sample wll be needed. It s crtcal that the entre sample stablty study be conducted wth a sngle lot of reagent and a sngle lot of calbrator. All of the samples studed and all of the QC materals must be analyzed usng the same lots of reagent and calbrator. It s also crtcal that the QC characterzaton runs and the sample stablty study be conducted wth, mnmally, the same lot of calbrator. Ideally, the same lot of reagent should also be used for the QC characterzaton runs and the sample stablty study, but ths may be more dffcult to organze. Samples wll be collected n multple cycles, startng on separate days, to smplfy the collecton process. Follow the collecton gudelnes descrbed n the Manufacturer s Specmen Collecton secton n the ndvdual protocol of nterest (e.g. total cholesterol, HDL cholesterol, LDL cholesterol, or trglycerde). See the Appendx of ths protocol for the concentraton dstrbuton for each analyte. On the frst day, collect blood from a subset of the total number of donors. For each ndvdual donor, harvest the serum and pool the serum f more than one tube was collected. Dvde the serum nto 4 alquots (or more f addtonal tme frames are to be nvestgated). Freeze 3 alquots at -70EC. Wthn 4 hours of collecton, analyze the fourth alquot n duplcate by the routne clncal method. At 7, 30, and 60 days, remove an alquot from the freezer and analyze t n duplcate by the routne clncal method. On addtonal days, collect blood from another group of donors. Repeat the protocol as on the frst day. Be sure to keep the samples from the multple collecton cycles separate to avod confuson. Durng the tme frame that the samples are frozen, run the method wth the QC materals every other day. Use the same lot of reagents and calbrator durng ths tme perod. A sample of a table that can be used to record data s ncluded wth ths protocol. After all samples have been analyzed, combne the data for the 7-days frozen from each collecton cycle. Evaluate the results by the t-test, comparng the 7-days frozen group to the fresh group. Lkewse, combne the data for the 30-days frozen from each collecton cycle and evaluate t usng the t-test. Fnally, combne the data for the 60-days frozen from each collecton cycle and evaluate t usng the t-test. CRMLN Sample Stablty Protocol October 2004 Page 3
t-test Perform the t-test for each nterval separately (e.g. 7-days frozen v. fresh). Calculate the average of the duplcate measurements performed for each sample and tme. average fresh = (R1 0 + R2 0 )/ 2 where R 1 0 and R2 0 are the replcates for the fresh sample. where R1 7 and R 2 7 average = (R1 + R2 )/ 2 7 7 7 are the replcates for the 7-days frozen sample. For each sample, calculate the dfference between the frozen alquot and the fresh alquot from the averages of the duplcates. dfference = average average 7 fresh For each tme frame (.e. 7-, 30-, and 60-days frozen), calculate the average dfference for all of the samples n the concentraton range. n dfference =1 avgdff 7days = n where n s the number of samples n the concentraton range. Calculate the SD dff for the pared dfferences. SD dff = n (dfference ) 2 n(avgdff 7days ) 2 =1 n 1 Use the pared t-statstc formula as follows: t = ( avgdff ) 7days SD dff n Use a table of crtcal values for t for a 2-taled " = 0.05 and degrees-of-freedom = n 1. If the value calculated for t s greater than the crtcal value, then there s a sgnfcant dfference between the frozen and fresh samples. CRMLN Sample Stablty Protocol October 2004 Page 4
Table 1: Sample Stablty Data Start Sample Fresh Results Frozen 7 Days Frozen 30 Days Frozen 60 Days Date ID Rep1 0 Rep2 0 Rep1 7 Rep2 7 Rep1 30 Rep2 30 Rep1 60 Rep2 60 CRMLN Sample Stablty Protocol October 2004 Page 5
Table 2: Qualty Control Lmts Record prevously determned QC lmts n ths table. Materal Mean 99% LCL 95% LCL 95% UCL 99% UCL 95% RL 99% RL LCL: lower control lmt UCL: upper control lmt RL: Range lmt CRMLN Sample Stablty Protocol October 2004 Page 6
Table 3: Qualty Control Results Materal 1 Materal Date Result #1 Result #2 Mean Range Mean n Control? Range n Control? CRMLN Sample Stablty Protocol October 2004 Page 7
Table 4: Qualty Control Results Materal 2 Materal Date Result #1 Result #2 Mean Range Mean n Control? Range n Control? CRMLN Sample Stablty Protocol October 2004 Page 8
Appendx: Concentraton dstrbuton These concentraton dstrbutons are based on usng 18 samples for the comparson whch assumes an analytcal CV of 1%. Total Cholesterol Dstrbute the samples usng the followng gudelnes: 20% samples from 120 to 180 mg/dl (3.10 to 4.67 mmol/l) 30% samples from 181 to 220 mg/dl (4.68 to 5.71 mmol/l) 30% samples from 221 to 260 mg/dl (5.72 to 6.74 mmol/l) 20% samples from 261 to 400 mg/dl (6.75 to 10.34 mmol/l). HDL Cholesterol Approxmately 60% of the samples should be dvded equally among each of the followng ranges. The remanng samples can fall nto any of the fve ranges; care should be taken to dstrbute the remanng samples over the concentraton range. Range 20-29 mg/dl (0.52-0.77 mmol/l) 30-39 mg/dl (0.78-1.03 mmol/l) 40-49 mg/dl (1.04-1.29 mmol/l) 50-59 mg/dl (1.30-1.55 mmol/l) 60-69 mg/dl (1.56-1.81 mmol/l) LDL Cholesterol Dstrbute the samples usng the followng gudelnes: 20% of samples < 100 mg/dl (2.59 mmol/l) 30% of samples from 100 to 130 mg/dl (2.59 to 3.36 mmol/l) 30% of samples from 131 to 160 mg/dl (3.37 to 4.14 mmol/l) 20% of samples from 161 to 400 mg/dl (4.15 to 10.35 mmol/l) Trglycerde Dstrbute the samples usng the followng gudelnes: 10% samples < 75 mg/dl (< 0.85 mmol/l) 25% samples from 75 to 124 mg/dl (0.85 to 1.40 mmol/l) 30% samples from 125 to 199 mg/dl (1.41 to 2.25 mmol/l) 25% samples from 200 to 299 mg/dl (2.26 to 3.38 mmol/l) 10% samples from 300 to 400 mg/dl (3.39 to 4.52 mmol/l) CRMLN Sample Stablty Protocol October 2004 Page 9