The Carcinogenic Potency Database is published in a two-sided, facing page, plot format. The plot includes information on many aspects of the 6540 experiments on 1547 chemicals that it summarizes. The following description of the plot, in conjunction with the Appendices, is intended as a guide for the reader to facilitate use of the data. A second format of the CPDB, which includes the same information is intended for viewing on a computer screen. There is a separate guide to the screen format. The screen version guide also describes each field in the format presented on the single web page for each chemical.
The published plot covers two facing pages and is organized alphabetically by chemical name. All experiments of a chemical are listed under the name of the test agent, and each experiment can be identified by a unique number in the plot.
This guide uses an example of one experiment from the CPDB plot in order to describe variables included in the plot, codes and conventions, and appendices. The example below is one experiment of phenolphthalein conducted by the NTP in female mice. The left side of the plot appears above the right side in the example but is on facing pages in the published plots. At the top of the example, as at the top of each page of the plot of the entire database, is a two-line header describing the type of information in each field. The header should be read across, alternating between the top and bottom line: first “Spe” then “Sex” then “Strain” etc. An experiment is defined as one sex of one species using a given route of exposure, from one research report. The codes in the CPDB are mnemonic for ease of use.
Immediately beneath the header in this example, we have inserted a set of numbers (1) - (30) which will be used in this guide to give details of the information in each field; this set of numbers does not appear in the plot.
Spe Strain Site Xpo+Xpt TD50 2Tailpvl
Sex Route Hist Notes DR AuOp
(1) (2)(3)(4)(5) (6) (7) (8)(9)(10) (11) (12) (13) (14)(15)16)
PHENOLPHTHALEIN 100ng..:..1ug....:..10.....:..100....:..1mg....:..10.....:..100....:..1g.....:..10
4958 M f b6c eat MXB MXB 24m24 : + : 508.mg Z P<.0005
a M f b6c eat --- mly 24m24 748.mg Z P<.003 c
b M f b6c eat --- lmt 24m24 1.61gm Z P<.003 c
c M f b6c eat ova MXA 24m24 1.90gm Z P<.003 c
d M f b6c eat --- hcs 24m24 4.08gm * P<.0005c
e M f b6c eat TBA MXB 24m24 1.74gm * P<.2
f M f b6c eat liv MXB 24m24 no dre P=1.
g M f b6c eat lun MXB 24m24 9.19gm * P<.3
RefNum LoConf UpConf Cntrl 1Dose 1Inc 2Dose 2Inc 3Dose 3Inc Citation or Pathology
Brkly Code
(17)(18)(19) (20) (21) (22) (23) (24) (25) (26) (27) (28) (29) (30)
PHENOLPHTHALEIN 77-09-8
4958 TR465 296.mg 1.53gm 15/50 388.mg 33/50 777.mg 40/50 (1.55gm 36/50) ---:hcs,lmt,mly; ova:sxb,sxs. C
a TR465 394.mg 4.88gm 15/50 388.mg 28/50 777.mg 33/50 (1.55gm 25/50)
b TR465 859.mg 8.28gm 1/50 388.mg 9/50 777.mg 10/50 (1.55gm 7/50)
c TR465 1.00gm 8.68gm 0/50 388.mg 7/50 777.mg 6/50 (1.55gm 5/50) ova:sxb,sxs.
d TR465 2.27gm 11.7gm 0/50 388.mg 2/50 777.mg 7/50 1.55gm 7/50
e TR465 642.mg n.s.s. 40/50 388.mg 39/50 777.mg 47/50 1.55gm 43/50
f TR465 3.96gm n.s.s. 21/50 388.mg 3/50 777.mg 6/50 (1.55gm 2/50) liv:hpa,hpb,hpc.
g TR465 2.56gm n.s.s. 6/50 388.mg 5/50 777.mg 6/50 1.55gm 8/50 lun:a/a,a/c.
For any given chemical, to determine whether results are reported in the CPDB, the chemical name can be searched in All Results for Each Chemical.
(1) The chemical name in capital letters is indicated under (1) in the top line for a set of experiments. In the plot, common synonyms and CAS number are also provided on the right side under (17). Also under the number (1), immediately below the chemical name, is the unique plot number for each experiment, i.e., one sex of one species from one research report. In this example, we use the line number 4958 which corresponds to the number in the plot of all CPDB. Each new number indicates a separate experiment. A set of lower case letters, in the example “a” through “g” identifies each tissue-tumor site. Only the site with the most potent TD50 value is plotted on the graph, i.e., one per experiment.
(2) The species used in this experiment is indicated under (2) and the column is headed by “Spe”. The letter “M” refers to mice, “R” to rats, “H” to hamsters, “D” to dogs, “N” to prosimians, and “P” to monkeys.
(3) The sex is indicated by “f” for female, “m” for male under (3). Occasionally an author in the published literature will report data only for both sexes together, and in these cases the code used is “b” for both.
(4) The strain or stock of animal is reported as a three-letter-code under strain (4). The Strain Appendix lists strain codes and definitions. Strains are coded just as they are referred to in the original publication. No attempt has been made to standardize the strain names; therefore, if different nomenclature is used by two authors who actually tested the same strain, then two different codes are used in the database. For monkeys and prosimians, this column is used for the species code, e.g., “rhe” for rhesus.
(5) The route of administration is indicated in the header line by “Route” and reported as a three-letter-code under (5). In the example, “eat” stands for administration in the diet. Route codes are listed in the Route Appendix and use mnemonic codes like “gav” for gavage.
(6 & 7) The site and histopathology are reported on this line under (6) and (7), and are marked in the header line by “Site” and “Hist”. Each is indicated by a three-letter-code, and the respective codes and definitions are provided in the Site and Histopathology appendices. Three-letter-codes have been created so that they are similar to the words they represent; for example, line 4958a reports “--- mly” which stands for “all sites, malignant lymphoma”.
For the NCI/NTP bioassays and the general literature, the nomenclature reflects the terminology used in the Technical Report or the published paper. The operational rule has been to retain what is published and not reinterpret or rename diagnostic categories. Thus, when various authors use different nomenclature for the same tissue or morphologic type of tumor, two different codes are used in the database. Occasionally it has been necessary to replace an adjective used for a tissue with a noun, e.g., the database uses kidney when renal is used in a paper. Some special considerations about the reporting of site and histopathology information from each source of data are as follows:
In the phenolphthalein example above, certain tissue and tumor codes are given in capital letters; these denote particular mixes of sites or tumor types from the NCI/NTP bioassays. (Capital letters are not used for papers in the published literature.) When these capitals appear, additional information about the specific pathology is presented on the right side of the plot under (29) where the header line reads “Citation or Pathology”. These special capitalized codes are used in the plot based on special mixes of tissue and tumor types from the NCI/NTP bioassays:
The mandatory sites are denoted by “MXB” in the tumor field (for “Mix Berkeley”) to indicate that the site was created especially for the CPDB and is not based upon the NCI/NTP evaluations in the Technical Reports. For every NCI/NTP experiment, the same mandatory sites are given per species: for mice, liv MXB (liver mandatory), lun MXB (lung mandatory) and TBA MXB (all tumor-bearing animals); for rats, liv MXB and TBA MXB.
For the NCI/NTP bioassays, these mandatory sites are always listed last in the experiment, in the order TBA, liv, and lun. Thus in the example given here, lines 4958e, 4958f, and 4958g, report for mice the TBA MXB, liv MXB, and then lun MXB. The specific pathology is given for liv MXB and lun MXB under (29) on the right side.
“MXA” (for “Mix Author”) is used to denote a combination of sites or tumor types which is taken directly from the Technical Report tables of primary tumors, and denotes a mix of tissues or tumors created in those tables. In the example, the site and histopathology for line 4958c, “ova MXA” are listed under (29). Whenever MXA appears on the left side, as in line 4958c “ova MXA”, the sites and/or histopathology which were combined are listed under (29) on the right side, in this case sxb (sex-cord stromal tumor, benign, bilateral) and sxs (sex-cord stromal tumor, benign).
“MXB MXB” denotes that a combination of tissues and tumors has been created by our group (MXB for “Mix Berkeley”), which consists of the aggregates of sites and histopathology evaluated in the Technical Report as “carcinogenic” or “clear” or “some” evidence of carcinogenic activity. In the example, under (29) for line 4958, MXB MXB is described as a combination of “---:hcs,lmt,mly; ova:sxb,sxs.” (This stands for all sites, histiocytic sarcoma; lymphoma malignant, thymic origin; lymphoma, malignant; ovary sex-cord stromal tumor, benign, bilateral; and ovary sex-cord stromal tumor, benign). See Berkeley Codes Appendix.
The site and pathology information from the literature experiments is given in the plot for individual tissues and tumors as it is for the NCI/NTP bioassays. It is usually not possible to combine sites from the published literature because, unlike the data available from the NCI/NTP bioassays, information is seldom reported about multiple tumor incidence in the same animal. When an author does give information about aggregated tissue or tumor types, the code “mix” is used in the plot to denote that specific sites and tumors are described in the paper. When the tumor types are not specified, the code “tum” is used. Mandatory sites from experiments in the literature are included in the database for the same tissues as the NCI/NTP bioassays. A TD50 is calculated for any mix of tumors reported in the mandatory site. All codes are in lower case letters. Whenever an author reported results for all animals with malignant tumors separately from all animals with benign tumors, we have included a TD50 for “tba mal” and “tba ben”.
(8 & 9) The exposure and experiment time are indicated in the header line by “Xpo+Xpt” under (8) and (9). Exposure time is the period over which the test agent is administered; if administration was 5 times a week for 40 weeks, for example, the exposure time is 40 weeks. Experiment time is the total time on test; it is not the age of the animals. It is measured from the start of the experiment to the time of death of the last dosed animal. Within a single experiment, all TD50s have the same exposure time and the same experiment time. Both times are always reported in the same units. When both are less than 100 weeks, exposure and experiment time are reported as “w” for weeks; when greater than this, “m” for months is used. For tests in long-lived experimental animals like dogs and nonhuman primates, “y” for years may also be used. When exposure time and experiment time are equal, then the duration of dosing was for the entire experiment. In the example, lines 4958a-4958g, the mice were dosed for 24 months, and the experiment lasted a total of 24 months.
(10) Notes, indicated in the header line by “Notes” under (10), provide additional information about the experiment in single-letter-codes which are defined in the Notecodes Appendix. This information is helpful in evaluating the experimental data. In the example, no notecodes are reported. Notecodes indicate such factors as: poor survival due to toxicity or disease (notecode “s”), variable or irregular dosing (notecode “v”), that the experiment was a serial sacrifice in a longer study (notecode “k”), or that histopathological examination was restricted to only a few tissues (notecode “r”). In some recent NTP bioassays, results for the kidney are reported in the Technical Reports for the standard histopathology protocol and separately for results including additional sections of the kidney. In such cases the word “with step” appears under (10), and a new line number is assigned in the plot for these results.
(11) The logarithmic scale under (11) is used for presenting the values of TD50 and its confidence limits (in units/kg body wt/day). The plot extends from 100 nanograms to 10 grams. On the scale itself, the location of 100 nanograms, 1 microgram (µg), 10 µg, etc., is indicated by underscoring; the points for 5, 50, 500 are denoted by a “:”. For each experiment, only the TD50 for the “most potent site” is plotted; this TD50 is listed first. For other sites within an experiment, the TD50 is not plotted, but all other information about it is given in the plot.
The “most potent site” is determined by ordering the TD50s in each experiment by statistical significance. If any TD50s are significant at the p<0.01 level, then these are listed first, in order of potency. Then follow all TD50s with p<0.10 sorted in order of potency. Last, all other TD50s are listed in order of potency. For the NCI/NTP bioassays, the mandatory sites are excluded from this sorted order, and are listed at the end in the order TBA MXB, liv MXB, and lun MXB, here as lines e, f and g. For literature experiments, we have excluded the category “tba” from this sorting of the target sites, and have listed it last.
In the example of phenolphthalein, there are five TD50s with statistical significance of p<0.01. The TD50 for MXB MXB is the most potent and thus appears first. The plotted point for this most potent site lies between 100 mg and 1 g on the scale, and the estimated TD50, 508 mg, appears in the column to the right of the scale under the header title “TD50”. These results indicate that 508 mg/kg body wt/day is estimated to halve the proportion of tumorless survivors at the end of a standard lifespan for female mice (in the absence of all other causes of death).
(12) The TD50 value on the plot is indicated by one of three symbols: “+” “±” or “>”, depending upon the p-value associated with the TD50 for the most potent site. In an experiment where the statistical significance of the most potent TD50 is p<0.01, the symbol “+” is used, and the plotted point is the most potent estimate of TD50, i.e., the smallest value of TD50 from among those with p<0.01, as in the example of phenolphthalein. In an experiment where the statistical significance is between 0.01 and 0.10 a “±” symbol is used to plot the most potent site. If there is no evidence for any statistical association, i.e., for all TD50s, p>0.10, then the symbol “>” appears just to the right of the 99% lower confidence limit. For these statistically nonsignificant experiments, the symbol “>” provides a lower bound for TD50, i.e., it is extremely unlikely that TD50 could be less than (more potent than) the value plotted.
A special symbol “<+” is used for cases where 100% of all dosed animals had the tumor(s) of interest and the TD50 was calculated with summary data; the “<+” appears at the upper confidence limit. With summary data, only an upper bound, but no TD50, can be estimated when all dosed animals have tumors.
For a few experiments, the symbol for the most potent TD50 is plotted with parentheses around it to indicate that the test did not meet our standard criteria; the exposure or experiment length usually was a week shorter than the inclusion rule, and the study was positive. The length criteria have been relaxed for some studies in nonhuman primates. (See the Species Appendix.)
Because there are both statistical and nonstatistical uncertainties in the estimation of TD50, we have calculated 99% confidence intervals for it. (See Statistical Methods for Estimating TD50.) Whenever the TD50 calculation is based on lifetable data, the symbol “:” denotes the confidence limits. When the TD50 calculation is based on summary data, i.e. for the general literature, the symbol “.” denotes the confidence limits.
In an experiment where the statistical significance of TD50 is p<0.01, both the lower and upper confidence limits are plotted. The calculated values for these intervals are presented on the right side of the plot under (20) and (21) where the header line reads “LoConf” and “UpConf” for lower confidence and upper confidence limits. In the example, the plotted confidence interval is 296 mg to 1.53 gm, indicating that the lower value would not reduce the proportion of tumorless survivors by half, while 1.53 gm/kg/body wt/day would more than halve it.
In an experiment where the statistical significance of TD50 is p>0.01, the 99% confidence interval will be open at the upper end. In such a case, the lower limit dose-rate would be unlikely to halve the proportion of tumor-free survivors, but no statement can be made about the carcinogenic effect of higher doses. In such cases, there is no “:” or “.” to the right of the TD50 on the plot.
When the plotted symbol is “<+” because 100% of the dosed animals had tumors and the TD50 calculation was based on summary data, the “<+” appears at the upper confidence limit. In such unusual cases the upper limit dose-rate would induce tumors in all animals, but the rate which would reduce the proportion of tumorless survivors by half cannot be estimated with summary data.
Occasionally, values of TD50 or confidence limits that should be plotted are too large to be printed within the range up to 10 grams; in such unusual cases no plotted symbol appears, but the value will be given in the field for TD50 (13) or confidence limits (20) or (21).
(13) The value of each TD50 is presented just to the right of the plot range (13), and includes the appropriate units (per kg) of body weight per day. The symbol “noTD50” appears instead of a numerical value whenever 100% of the dosed animals had the tumor(s) of interest and the TD50 was calculated with summary data. The symbol “no dre”, for no dose-related effect, indicates that TD50 is not estimable for some other reason. For statistically non-significant TD50s the numerical value may be impossibly large.
(14) The shape of the dose-response curve for each TD50 appears in (14) under the header "DR". The codes and definitions are listed in the Dose-Response Curve and Plot Symbols Appendix. The shape of the dose-response has been determined by a test for departure from linearity. (See Statistical Methods for Estimating TD50.) If there was no significant departure from a linear dose-response, then the curve shape is listed as linear, and the symbol “*” appears. For experiments with three groups of animals including controls, a significant departure from linearity with upward curvature is denoted by the symbol “/”. If there was a significant departure from linearity with downward curvature, then the TD50 is calculated without the data from the highest dose group, and the symbol “\” appears. We have adopted this convention to obtain the best estimate of TD50 by using only the linear portion of the dose-response curve in the calculation (Gold et al., 1986). The groups that are excluded from the reported TD50 calculation are indicated by parentheses around the tumor incidence data under (27) and (28).
When there are more than three dose groups (including controls) in the TD50 calculation and there is a significant departure from linearity, the symbol “Z” is listed under (14). When there is a blank space for the shape of the dose-response, there are two possible reasons. First, there may be only one dose group and a control group in the experiment, in which case there is not enough information to determine a curve shape. Second, there may be no dose-related effect, in which case the code “no dre” appears in (13) and “P=1.” appears in (15).
(15) The two-tailed p-value appears in (15), under the header “2Tailpvl”. (See Statistical Methods for Estimating TD50.) This value indicates the statistical significance associated with testing whether the slope of the dose-response is different from zero. All values are given to one significant figure. When there is no dose-related effect or the slope is negative, then “P=1.” appears in (15). The lowest p-value reported is p<0.0005. Note that the significance level listed in this column determines which symbol will be plotted for the most potent site in each experiment.
(16) The opinion of the original author, as to the tumorigenicity of the test agent at the site for which the TD50 was calculated, is given in the last column on the left side of the plot (16) under the heading “AuOp” for Author’s Opinion. Our rule for reporting opinions from all sources of data has been to record all clearly stated evaluations of tumorigenicity at the site(s) included in a TD50 calculation. (See the Author’s Opinion Appendix.)
Our conventions for reporting the author’s opinions from the NCI/NTP Technical Reports are based upon the text of the Report and the statistical analysis tables. An author’s opinion is listed for all TD50s except: Berkeley Mixes (MXB) and the statistical sites, i.e., those included in the tables but not considered evidence for carcinogenicity in the text of the Technical Report. For these cases, the opinion column is blank, as in lines 4958, 4958e, f, and g of the phenolphthalein example.
A “c” in the author’s opinion column indicates that the text of the Report stated that at the site on which TD50 is based, the compound was carcinogenic under the conditions of the bioassay for NCI Technical Reports or the Abstract indicated that there was “clear evidence of carcinogenic activity” for NTP Technical Reports. See for example, the opinion column in the example of phenolphthalein for lines 4958a, b, c, and d. In NCI Technical Reports an “a” indicates an opinion that the incidence of tumors at that site(s) was evaluated as associated with administration of the compound under the conditions of the bioassay, or that the evidence for carcinogenicity was suggestive; these opinion codes in the CPDB are consistent with updated evaluations by NTP (Haseman et al., Environ. Health Perspect. 74: 229-235, 1987). For all NTP Technical Reports that use the current evaluation methodology, we report all sites listed in the summary table in the Abstract with the opinion for carcinogenic activity: “c” for clear, “p” for some, and “e” for equivocal. (See the Author’s Opinion Appendix.) The “a” opinion for NCI generally corresponds to an opinion of “e” for NTP.
The symbol “–” will appear in the opinion column for the most potent site in an NCI/NTP bioassay to denote the evaluation that the compound was not carcinogenic in that sex of that species under the conditions of the bioassay. In most cases, the “–” appears for the TD50 calculated for TBA, which is our convention whenever there is no evidence for carcinogenicity. For experiments evaluated as inadequate by NCI or NTP in the Technical Report, there is an “i” in the opinion column for the most potent site (See the Author’s Opinion Appendix.).
There are some cases when the “–” appears for a statistical site, i.e., one not evaluated as evidence for carcinogenicity in the Report, but which was statistically significant according to the statistical tables in the Report, and also had a TD50 significance level of p<0.05. When this TD50 is the only evidence for a treatment-related effect, and thus the most potent site, we have indicated this by placing a “–” in the opinion column and flagging the TD50 with a # sign in the plot just to the left of the TD50 value. See for example, “acetohexamide,” for male rats, line 35. Note that the p-value associated with the TD50 determines the symbol plotted for the TD50; this is independent of the author’s opinion.
For bioassays in which some target sites were evaluated as treatment-related, the statistical sites are also reported but the opinion column is left blank. In order to make it clear that NCI/NTP did not evaluate these statistical sites as evidence of carcinogenicity, we put an “S” for statistical in the last column on the right side of the plot (30), under the header “Brkly Code.” (See the Berkeley Codes Appendix.)
In the general literature whenever the author evaluated the proportion of animals with tumors at a particular site as treatment-related, a “+” will appear in the opinion column (16) for the TD50 calculated for that site. Such stated opinions as “positive,” “carcinogenic,” “induced,” “treatment-related,” and “tumorigenic,” fit this category. The symbol “+” will only appear in the opinion column for a tissue-tumor combination in one of the following two cases: (1) the author gave a positive opinion for the particular target sites included in a tissue-tumor combination, or (2) the occasional case where an author evaluated the compound as carcinogenic without specifying the target site, and we have indicated this with a “+” symbol in the opinion column for the category “all tumor bearing animals (tba)”, e.g. line 4621 for female rats in the CPDB plot for N-nitrosopiperazine.
Similarly, the opinion column will contain a “–” only when either (1) the author stated an opinion that there was no carcinogenic effect at the particular sites, or (2) the author concluded that there was no treatment-related effect in the experiment, in which case sites reported have a “–” in the opinion column.
Sites which an author did not evaluate as positive are included in the database only when the statistical significance associated with an increased percentage of dosed animals with tumors is p<0.05 (chi-square, one-sided p-value), or when the tissue is a mandatory site.
When no opinion about carcinogenicity is stated in the published paper for sites which are reported in the plot, the author’s opinion column is left blank. This may occur either for mandatory sites, or for included sites which were not unequivocally evaluated by the author. If additional information about evaluations was obtained directly from the author, then “pers.comm.” appears after the citation on the right side of the plot under (29).
In summary, the symbol for the author’s opinion column in the general literature reflects what the author actually stated. Sites evaluated as positive are given a “+”. Sites evaluated as negative are given a “–”. The symbol “+” is used for tba when the compound was evaluated as positive, and no specific target site was evaluated as positive. For all other opinions the author’s opinion column is blank.
Occasionally, when there is a “c”, “p”, “a”, or “e” in the opinion column for an NCI/NTP bioassay, or a “+” for a test from the general literature, the positive evaluation was made because the incidence among dosed animals was high in comparison to historical control incidences; this occurs, for example, when there is a rare tumor among dosed animals. The actual numbers of animals bearing such tumors may be quite low, thus making the estimate of TD50 unreliable. In such cases, we have indicated that the author’s opinion was based on historical control comparisons by putting “+hist” to the left of the TD50 value. See, for example, 5-nitro-o-toluidine for female mice (line 4300b).
(17 & 18). The plot example of phenolphthalein continues on the right side, by first repeating the identification numbers given for each line on the left side of the output, e.g., 4958, 4958a, 4958b etc. In (17) the chemical name is also repeated, followed by common synonyms, in some cases. In the example, there are no synonyms reported. The Chemical Abstracts Service registry number (CAS) is reported under (18); in the example, 77-09-8. To find a chemical in the CPDB by its CAS number, see the Synonym Appendix.
(19) Under the header “RefNum” in (19), is the unique reference number assigned to each paper in the database. For NCI/NTP bioassays, this is the Technical Report number, TR465 in the example. For literature tests the number is a chronologically assigned paper identification number for the CPDB.
(20 & 21) The lower and upper confidence limits for each TD50 are presented in (20) “LoConf” and (21) “UpConf” respectively. When the abbreviation “n.s.s.” appears for either the lower or upper confidence limit, it denotes not statistically significant. Whenever the statistical significance of TD50 is p>0.01, then the upper 99% confidence limit will not be calculated. See, for example, line 4958e in the phenolphthalein example. When the lower confidence limit is “n.s.s.” this usually indicates that there were no tumors or only one tumor of the specified type in the experiment, and the lower confidence limit was not estimable; most often this occurs for mandatory sites. Occasionally the n.s.s. occurs for the lower confidence limit because 100% of dosed animals had the tumor(s) of interest and hence no lower confidence limit could be estimated with summary data. This is also indicated by the plot symbol “<+”.
(22-28) Beginning in (22) on the right side, and extending through (28), we report the proportion of animals with tumors and the average dose rate in mg/kg body wt/day which we have calculated for each dose group in the experiment. The proportion of animals with tumors for TD50s which have been calculated with lifetable data are presented here in summary form, i.e., the number of animals with tumors by the end of the experiment. The denominator reported for NCI/NTP bioassays is the starting number in each group.
Reading across (22)-(28), under “Cntrl” (22), we list the proportion of control animals with the tumor types in the TD50 calculation; the average daily dose in the lowest dose group is given under “1Dose” (23); the proportion of animals in that group with the tumor(s) is listed under “1Inc”, for incidence, (24); the next highest average dose “2Dose” (25) and the proportion of those animals with the tumor(s) “2Inc” (26) is given next, etc., for as many dose groups as there are in the experiment.
Whenever the TD50 was calculated without the data from the highest dose group(s) (i.e., there was a significant downward departure from linearity), we have indicated this fact with parentheses around the data which were omitted from the final calculation. (See Statistical Methods for Estimating TD50.) Thus, whenever the shape of the dose-response on the left side of the plot is “\” under (14), the parentheses appear on the right side around the appropriate data. In the example, line 4958 has one dose omitted from the final calculation. Whenever there were more than three groups (including controls), and the dose-response was nonlinear, there is a “Z” under (14) on the left side. If the departure from linearity in such cases was downward, then this fact is indicated by parentheses around the group that was excluded from the TD50 calculation, as is the case for lines 4958, a, b, and c. If there are no parentheses and a “Z” appears in the curve-shape column, then the dose-response had at least 3 doses and a control and the dose-response curved upward, e.g. “butylated hydroxyanisole” in male rats, line number 933.
For the proportion of animals with tumors, the number of animals reported in each group for NCI/NTP bioassays is the number at the start of the experiment; the TD50 was estimated with lifetable data. In the example, there were 50 in all groups. For some early NCI experiments pooled controls were used in evaluating evidence for carcinogenicity in the Technical Report, and we have calculated TD50s with those pools as well as the matched controls. Data using the pooled controls are indicated by the letter “p” following the control incidence in column (22), by the word “pool” on the left side of the plot following the Notes in (10), and by assigning a different line (experiment) number to the pooled data. See, for example, “aldrin” for male mice on line 241.
The proportion of animals with tumors presented here is the number of animals used in the TD50 calculation. Many authors have reported only the starting number of animals. Whenever the published paper had additional information, i.e., the number of animals alive at the time of appearance of the first tumor, or if that was not reported the number examined histologically at the site, then this number is used in the denominator of the proportion of animals with tumor. This is a more accurate description of the number of animals at risk of tumor. These data were used in the TD50 calculation. In these cases, the notecode “e” for effective number appears on the left side of the plot under “Notes” in (10). Otherwise, the data reflect the number of animals started in each group. Since experimental designs vary in the literature, the incidence data may include a control and only one dose group or perhaps, a control and several dose groups. We correspond with about half the researchers about their published papers and are sometimes able to report tumor incidence data that are not given in the publication but improve the estimate of the TD50, e.g., number of animals alive at the first tumor in the denominator.
For the general literature, use of pooled control data in the CPDB is rare and is only done when matched data were not reported (only a few tests, and before 1985); these are indicated on the plot in the same way as for NCI/NTP bioassays. Whenever vehicle control data were available, either in NCI/NTP or general literature, these were used in the CPDB.
(29) Reading across the right side of the plot, under “Citation or Pathology” (29) for the NCI/NTP bioassays, we present the three-letter-codes for all sites and histopathology included in TD50s which are “Author’s Mix” (MXA) or “Berkeley Mix” (MXB). This includes the MXB mandatory liver and lung sites, our combinations of sites which were individually evaluated as positive in the Report (MXB), the statistical sites which included more than one site in the Tables of Analyses (MXA), and any combination of sites evaluated as treatment-related in the Technical Report (MXA). The three-letter-code for each tissue in the TD50 calculation is followed by a “:” and then by the three-letter-codes for each category of neoplasm included in the calculation. A “.” follows the last three letter tumor code in each mix. The definitions for these codes are given in the Site and Histopathology appendices.
For the published literature, a citation is provided, listing the first author, code for the journal or book title, volume number, pages, and year of publication. The full titles of the four-letter-codes for the names of references are listed in the Journal Code Appendix. The abbreviation pers.comm. indicates that additional data for the TD50 calculation or author’s opinion were obtained through personal communication with the author(s). On the plot a new citation for a published paper is listed whenever experimental results are from a different paper, i.e. if several experiments are reported consecutively in the CPDB from a single paper, then the citation will not repeat. When the next experiments are from a different paper, the new citation will be reported. If additional experiments are later reported from an earlier citation, the citation will then be repeated. If pathology codes appear under (29), results are for an NCI/NTP bioassay.
A complete bibliography of all papers in the CPDB is given in two appendices: Part 1 lists the NCI/NTP Technical Reports by chemical name, and indicates Technical Report number and year of publication. Part 2 gives complete citations to articles, books, and reports in the general literature.
(30) The last column of the plot, is used only for NCI/NTP bioassays. Under the header “Brkly Code”, we indicate that a TD50 has been included in the database because of a decision by the Carcinogenic Potency Project (Berkeley) rather than because the sites were evaluated as treatment-related or combined in the NCI/NTP Technical Report. (See the Berkeley Code Appendix.)
The capital letters “C”, “M” and “P” are used in the Berkeley Code column for Berkeley Mixes (MXB). The letter “C” denotes a TD50 calculated for the combination of all sites evaluated in the Technical Report as clear evidence for carcinogenicity in this experiment. See line 4958 in the phenolphthalein example. The letter “C” denotes a combination of tumors evaluated as clear evidence. The letter “P” indicates a combination of tumors evaluated as some evidence. The letter “M”, for mix, is uncommon and denotes a combination of all sites evaluated as “C” or “P” in a bioassay where “clear” and “some” evidence were evaluations for different target sites in the same experiment. The letter “S” is not a MXB, but rather indicates that the TD50 has been included in the plot because the sites were statistically significant in the tables of the Technical Report and the TD50 was significant at the p<0.05 level; however, NCI/NTP did not evaluate the site as evidence of carcinogenicity. For all mandatory sites, the column for “Brkly Code” is blank.
In the plot, bioassays of a particular chemical are organized under the chemical name, and these names are ordered alphabetically. Within each compound, the experiments are ordered alphabetically by species code, so that dogs would appear first, then hamsters, mice, prosimians, monkeys, and finally rats. Within a species, the bioassays are ordered by the code for the strain or stock. If there is an NCI/NTP bioassay of the chemical, then all experiments using that strain are reported first, followed by the strain used in any other experiments providing lifetable data, and finally by any remaining strains ordered alphabetically. Within the strain, the bioassays of females are reported first. Thus, when there is an NCI/NTP bioassay, “b6c” mice will appear first, and all experiments using “b6c” female mice would be reported before any experiments using “b6c” males.
To facilitate easy reference back to this guide when using the large plot, we have described each variable here in the order in which it appears in the plot. The titles in the header of the large plot also refer back to the categories in this guide. Abbreviations and symbols are defined in detail in the Appendices.
Return to the Carcinogenic Potency Project Home Page:
Last updated: August 6, 2007