use of local vs. central laboratories in analysis and handling of the data will also
... The most common categories for laboratory tests for human clinical trials are ...
NESUG 2007
And Now, Presenting...
Presentation and Handling of Clinical Laboratory Data From Test Tube to Table Randall K. Carlson, Wilmington, DE and Nate Freimark, Lakewood, NJ Omnicare Clinical Research. Inc. INTRODUCTION In human clinical trials of new pharmaceutical agents, clinical laboratory data is a key element of the safety profile for the drug agent being studied. In this paper, background about the categories and types of laboratory parameters commonly collected in clinical trials will be presented. The differences and the various challenges surrounding the use of local vs. central laboratories in analysis and handling of the data will also be discussed. The need for unit conversion and information on the various SAS® methods used to accomplish this will be outlined. Finally the paper will provide several examples of typical summary tables commonly used to present clinical laboratory results. These include descriptive statistic, shift tables, CTC grading summaries, as well as examples of the types of graphs that can be produced using SAS/GRAPH®.
BLOOD COMPONENTS AND COMPOSITION Human Blood is made up of about 45 percent cellular components (primarily Red Blood Cells (RBC), White Blood Cells (WBC), and platelets) and 55 percent fluid (Plasma). The table below lists the major components and their primary biological function
COMPONENTS
FUNCTION
RED BLOOD CELLS
CARRY OXYGEN
WHITE BLOOD CELLS
FIGHT INFECTION
PLATELETS
CLOTTING
PLASMA
TRANSPORT NUTRIENTS, HORMONES, PROTEINS
CATEGORIES OF COMMON LABORATORY TESTS The most common categories for laboratory tests for human clinical trials are Hematology, Blood (or Serum) Chemistry, Urinalysis and Coagulation. Less common categories are Microbiology, Stool specimens, and testing for specific drugs. While the grouping specific tests sometimes varies from case report form to case report form, in general the Hematology category is of those tests involving the cellular components of the blood, while Blood Chemistry focuses on the plasma components. The Urinalysis group are those tests involving components of the Urine. Coagulation Tests are those that involve blood clotting, these are sometimes grouped into the Hematology test category. Microbiology tests are those testing for the presence of specific bacteria, and fungi.
COMMON LABORATORY TESTS The determination of what tests are actually collected during a clinical trial in some degree is determined by the drug being studied and is generally defined in the protocol. Below is a list of commonly collected laboratory tests grouped by category. The groupings are not hard and fast and different studies sometimes group the lab tests in different groups. For example, combining coagulation and hematology into one group, is a common practice. The grouping of lab tests is often used not only for CRF collection, but also to group together the test for presentation in 1
NESUG 2007
And Now, Presenting...
the study results.
HEMATOLOGY RBC (Red Blood Cell Count) Hemoglobin Hematocrit MCV (Mean Corpuscular Volume) MCH (Mean Corpuscular Hemoglobin) MCHC (Mean Corpuscular Hemoglobin Concentration) WBC (White Blood Cell Counts) Neutrophils (% and/or absolute) Lymphocytes (% and/or absolute) Monocytes (% and/or absolute) Eosinophils (% and/or absolute) Basophils (% and/or absolute)
BLOOD CHEMISTRY Albumin LDH (Lactate Dihydrogenase) Alkaline Phosphatase Phosphorus Bicarbonate Potassium BUN (Blood Urea Nitrogen) SGPT (ALT - alanine transaminase) Calcium SGOT (AST- aspartate transaminase) Cholesterol Sodium Creatinine Total Bilirubin Chloride Total Protein Calcium Triglycerides GGT Uric Acid Glucose
COAGULATION Prothrombin Time (PT) Partial Thromboplastin Time (PTT) International Normalization Ratio (INR)
URINALYSIS Urine pH Urine Specific Gravity
2
NESUG 2007
And Now, Presenting...
FUNCTIONAL TESTS Certain Blood chemistry tests are often lumped together because they are indicators for organ and biological functions. They are most often done to make sure that no organ or biological system is being adversely affected by the drug being studied. In some studies these values are reported as a group to add in reviewing the functional data. Some of the common groupings are in the table below
System or Organ Liver Kidney Pancreas Electrolytes Nutritional Lipids
Lab tests ALT, AST, Alkaline Phosphotase, GGT, LDH, Albumin, Bilirubin BUN, creatinine Amylase Sodium, Potassium, chloride Glucose, fats Triglycerides, Cholesterol(LDL, HDL)
CENTRAL VS LOCAL LABS The laboratory data collected for clinical trials is typically handled by two types of labs. Central Labs, such as Covance, CCLS, and Quintiles and local labs which are general on-site in the hospital, or medical unit where the patient visit is taking place. Each type of lab has advantages or problems associated with it. Central labs have the advantage in that each test will be conducted using a standard procedure, which makes for better direct comparison among patients. Data is usually provided electronically. In most cases the central lab will also apply the normal ranges to each patient and lab test, as well as provide the results in the units specified by the sponsor or at least in a common unit based on their own defaults. But of course the samples need to be sent to the central lab which means there is a delay in results to the investigator. Local labs are most often used when an immediate result is needed to allow dosing or entry into the trial. The data is usually captured directly on the case report form. Local labs also complicated by the fact that you have to collect normal ranges from each lab, and sometimes they are not available. The unit of collection is dependent on the lab and will often vary from lab to lab. Standardization of the results to a single unit is often required.
UNITS OF MEASUREMENTS The units of measurement for a given lab test often very from lab to lab, in part based on the country the lab is located in. The labs in the US often use what is referred to as Conventional Units, while international labs use SI (Systeme Internationale). SI units are usually preferred for international trials, while US only trials often make use of the conventional units. In some cases the tests are expressed and analysis in both units.
PREPERATION FOR ANALYSIS Prior to analysis, it’s important that all the values in a given lab test are expressed in terms of a single unit in order to compare “apples” to “apples”. It is also important that that lab test be labeled with a single Parameter Name, so that all the “apples” are included. . There are several ways to program the standardization of the lab parameter, one is to use assignments statements, one for each of the different lab parameter names observed. If labvar = “ALBU’ then lbtest = ‘ALBUMIN’; else if labvar ‘ALBUMIN” then lbtest = ‘ALBUMIN’; else if labvar = ‘ALKPH’ then lbtest = ‘ALkALINE PHOSPHATASE’; else if labvar = ‘ALKPHOS’ then lbtest = ‘ALKALINE PHOSPHATASE’;
3
NESUG 2007
And Now, Presenting...
Another useful method to assign lab names is to create a format and use a put statement. proc format; value $lbtest 'ACANTHOCYTES'='ACANTHOCYTES' 'ALBU','ALBUMIN'='ALBUMIN' 'ALKPH','ALKALINE PHOSPHATASE'='ALKALINE PHOSPHATASE‘ ; lbtest = put(labvar,$lbtest.);
Once the lab parameter name is standardized the values for that lab test must all be expressed as a single unit. A number of methods were presented in a paper at NESUG 2006a. As an example an approach is shown below. First the units are changed and an appropriate conversion factor is assigned. Finally the conversion is applied to the result.
select(lbtest); when('BASOPHILS, ABS') do; select(labunit); when('K/UL','10^9/L') do; factor=1; lbstresu='10^9/L'; end; otherwise factor=.; end; end; lbstresn=lborresn*factor;
TYPICAL LAB SUMMARIES Lab summary tables will vary from study to study depending on what the objectives of that particular clinical trial, Of the more common types are table displaying descriptive statistics such as TABLE 1. They usually describe the lab value and its change or percent change from baseline at various visits or time points. The most common statistics are usually the mean, median, standard deviation, minimum and maximum values. This table would then be repeated for any lab tests collected. When there is a large variation in the results that makes direct comparison difficult, such as cases where local labs are used, shift tables are often used. The lab values are classified as above the normal range, within the normal range, or below the normal range. And then these values are summarized based on the change of categories between baseline and some subsequent time point (usually the last visit in the study). An example of this type of table is shown in Table 2. Another common type of table is to categorize the values into clinically significant categories, such as below a predefined critical value. Then the number of patients who fall into the categories are summarized. Table 3 is an example of this type of table. A very specific version of this type of table is when the Lab values are grouped based on the standard CTC toxicity grading. Table 4 is an example of this. 4
NESUG 2007
And Now, Presenting...
GRAPHICAL SUMMARIES Laboratory results are good candidates for graphical summarization. As a graph can give you a quick picture of increases and decrease in lab values over time, as well as a comparison between treatment groups. They are also will reveal data points which are outliers that bear further investigation. There are almost as many examples of graphs as there are studies, as they are often optimized to tell a specific story for a particular study, so it is not possible in the scope of this paper to present them all. One of the more common types of graphs is the box and whisker plot. Figure 1 is an example of this type of plot. In this case it is plotting the value of one lab parameter by visit for a specific treatment group. The line in the middle of the box represents the median value. The box itself are the values representing the 25th and 75 percentile and the whiskers represent the values within 1.5 of the interquartile ranges. Anything falling outside that range is plotted as a point using the symbol specified in the value option. The BOXT can also have values from BOXT00 to BOXT25. this will change the limits on the top and bottom whiskers. This type of plot allows for comparison of the median values over time, as well as show the variability at a given visit. axis1 label=( angle=90 "&PARAM") ℴ axis2 label=('WEEK OF STUDY') offset=(0.5 in); symbol1 interpol=boxt value=x; proc gplot data=plotds; plot &var*visn / noframe vaxis=axis1 haxis=axis2
Figure 2 is an example of another type of common plot, where two or more treatment groups are compared. In this case the mean of the lab test is being is being plotted for each visit, as well as displaying the values within +/1 standard deviation. To accomplish this plot, the mean and standard deviation are first determined at each visit for each treatment group. A data set is created containing the values for mean and then values for the MEAN-SD*1 and MEAN+SD*1 are calculated for each visit and treatment group.
proc univariate data=deff_n noprint; by parm trtname visit; var lbstresn; output out=deff_v n=n1 mean=mean1 std=sd1;
data deff_v; set deff_v; by parm trtname visit; val=mean1; output; val=mean1+sd1; output; val=mean1-sd1; output;
5
NESUG 2007
And Now, Presenting...
This data can then be plotted using the hilo option of the symbol statement. You can vary the options in the symbol statement to change the symbol displayed and the line used to join the mean values for each treatment group, so that they display differently which aids in identifying the data points. The legend statement also controls how the terms are displayed in the legend box.
symbol1 l=1 value=diamond width=3 interpol=hiloj line=1; symbol2 l=1 value=square width=3 interpol=hiloj line=5 ;
legend1 across=1 cborder=black label=none position=(top inside center) across=2 value=("TRTA(1-4 MG QID)" "&ztrtb&ztrtbb") mode=share;
proc gplot data=ps nocache; plot val*window=trtname / vaxis=axis1 haxis=axis2 legend=legend1;
6
NESUG 2007
And Now, Presenting...
Table 1 - Descriptive Statistics Serum Chemistry: ALT (U/L) Analysis Population: All-treated population ____________________________________________________________________ Pooled ______________(N=679)______________ Observed Change from BL ____________________________________________________________________ Baseline Mean, SD 56.3 , 140.77 Median 30.0 Min, Max 4 , 2852 n 571 End-of-Therapy Mean, SD Median Min, Max n
54.6 , 33.0 5 , 544
Post-drug follow-up Mean, SD Median Min, Max n
41.6 , 28.0 5 , 447
73.11 840
43.11 318
1.9 , 2.0 -662 , 475
-10.9 , -1.5 -767 , 392
80.82 445
84.97 192
____________________________________________________________________
7
8 2( 4( 0 0
0 0 0 0
0 0 0 0
1.6%) 3.1%)
3.9%)
4.7%)
3.1%)
4.7%)
2( 1.6%) 101( 78.9%) 0 1( 0.8%)
0 6( 0 5(
0 6( 0 4(
0 0 0 0
0 0 0 0
0 0 0 1(
0.8%)
0 18( 14.1%) 0 0
0 5( 3.9%) 0 112( 87.5%)
0 4( 3.1%) 1( 0.8%) 112( 87.5%)
Low 1( 0.8%) 1( 0.8%) 0 2( 1.6%) Normal 4( 3.1%) 103( 80.5%) 0 16( 12.5%) High 0 0 0 0 Missing 0 1( 0.8%) 0 0 _______________________________________________________________________________________________________
Low Normal High Missing
HEMATOCRIT (%)
HEMOGLOBIN (GM/DL)
Low Normal High Missing
Low Normal High Missing
EOSINOPHILS (%)
BASOPHILS (%)
Laboratory Parameter Screening N(%) N(%) N(%) N(%) ______________________________________________________________________________________________________
___________________________________________________________________
Laboratory Hematology Test Value Shifts Relative to Normal Reference Range Phase III Studies xxx Intent-to-Treat Population Treatment Group=Placebo (N=128) ______________________________________________________________________________________________________ Final Low Normal High Missing
Table 2 – Shift table
NESUG 2007 And Now, Presenting...
NESUG 2007
And Now, Presenting...
Table 3 – Clinically Significant Phase III Studies xxx Intent-to-Treat Population ___________________________________________________________________________________________ Number(%) Subjects ____________________________________________________________________________________________ Placebo Comparator (N=128) (N=127) Lab Parameter (Units) _________________________________________________________________________________________ Abnormality Criteria Screening Final Screening Final _________________________________________________________________________________________ HEMATOCRIT (%) male: