Selection8528

Allocation (randomization): Was administered dose or exposure level adequately randomized?14019

Randomization requires that each human subject or animal had an equal chance of being assigned to any study group including controls (e.g., use of random number table or computer-generated randomization). This applies to a concurrent negative control group (i.e., a group for which exposure is to vehicle or sham exposure or un-treated / cage controls) which must be included in the study to address randomization as well as any positive control group that may be part of the study.

PROMPTING QUESTIONS

For each study:

• Did each animal or litter have an equal chance of being assigned to any experimental group (i.e., random allocation)?
• Is the allocation method described?
• Aside from randomization, were any steps taken to balance variables across experimental groups during allocation?

Allocation (blinding): Was allocation to study groups adequately concealed at the start of the study?14020

Allocation concealment requires that research personnel do not know which administered dose or exposure level is assigned at the start of a study.

Allocation concealment prior to assigning the exposure level or treatment group (along with randomization in question #1) helps to assure that treatment is not given selectively based on potential differences in human subjects or non-human experimental animals. Allocation concealment requires that research personnel allocating subjects or animals to treatment groups (including the control group) could not foresee which administered dose or exposure level is going to be assigned at the start of a study.

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the study report blinding or other methods/procedures for reducing observational bias?
• If not, did the study use a design or approach for which such procedures can be inferred?
• What is the expected impact of failure to implement (or report implementation) of these methods/procedures on results?

Note:

1. a question under performance bias addresses blinding of personnel to treatment during the study;
2. a question under detection bias addresses blinding of outcome assessors.

---

Performance8530

Were experimental conditions identical across study groups?14023

Housing conditions and husbandry practices should be identical across control and experimental groups because these variables may impact the outcome of interest. Identical conditions include use of the same vehicle in control and experimental animals and the rating for this risk-of-bias element will depend largely on the consistent use vehicle across treatment groups. That is because under current reporting practices it is unlikely that similarity of conditions will be explicitly reported in most animal studies. Thus, we will assume unless stated otherwise that experimental conditions (other than treatment vehicle) were identical across groups.

Were the research personnel blinded to the study group during the study?14024

Blinding requires that study scientists do not know which administered dose or exposure level the animal is being given (i.e., study group).

---

Detection8532

Can we be confident in the exposure characterization?14026

Confidence requires valid, reliable, and sensitive methods to measure exposure applied consistently across groups.

It is essential that these criteria are considered, and potentially refined, by assessment teams, as the specific variables of concern can vary by exposure (e.g., reverberation chamber versus exposure coming from a cell phone).

PROMPTING QUESTIONS

·      Were body/tissue internal exposure metrics measured or calculated for the particular conditions in the study (SAR values (W/kg), SA (J/kg) induced electric field in (V/m), internal magnetic field strength in H/m or equivalent unit, respectively?

Can we be confident that possible body temperature increases have been properly assessed in the animals?14083

Did the study provide data on the body temperature of the animals?

PROMPTING QUESTIONS

For each study:

• Did the authors take steps to measure or calculate body temperature in the animals (i.e. in a pilot study)?
• Is the exposure very low and body temperature increases will be negligible (to be a risk of bias)?
• Are there concerns about significant increases in body temperature?

NOTE: There are no studies reporting that increases of body temperature by 1°C lead to cancer (There is limited evidence in experimental animals for the carcinogenicity of very hot water at 65 C or above, IARC vol 116), whereas other biological reactions and certain biomarkers can be changed due to temperature changes of 1°C (but none of these have been established as mechanisms of carcinogenicity).

Can we be confident in the outcome assessment?14027

Confidence requires valid, reliable, and sensitive methods to assess the outcome and the methods should be applied consistently across groups.

Detection bias can be minimized by using valid and reliable methods to assess the outcome applied consistently across groups (i.e., under the same method and time-frame). The objectivity of procedures used for measuring and reporting an outcome will impact the degree to which outcome assessors could bias the reported results.

NOTE: blinding of pathologists is not required nor common in academia, government agencies, industry and is not be considered as risk of bias. The initial review of slides is conducted unblinded (not masked), the secondary review is blinded (or masked) (Sills et al., 2019). This procedure is consistent with the position of the Society of Toxicologic Pathologists (STP) and the American College of Veterinary Pathologists (ACVP) (Crissman et al., 2004).

---

Attrition8531

Were outcome data complete without attrition or exclusion from analysis?14025

Attrition rates are required to be similar and uniformly low across groups with respect to withdrawal or exclusion from analysis.

Attrition or exclusion bias refers to systematic differences in the loss or exclusion from analyses of participants or animals from the study and how they were accounted for in the results.

PROMPTING QUESTIONS

For each study:

• Are all animals accounted for in the results?
• If there are discrepancies, do authors provide an explanation (e.g. death or unscheduled sacrifice during the study)?

If unexplained results omissions and/or attrition are identified, what is the expected impact on the interpretation of the results?

---

Selective Reporting8533

Were all measured outcomes reported?14028

Selective reporting is present if pre-specified outcomes are not reported or incompletely reported.

The reporting of the outcomes in this metric does not consider the quality of the data presentation, e.g., tranparency, description of the data (SEM or raw data).

Are the results and analysis presented in a in a way that makes the data usable and transparent?14084

CORE QUESTION

Are the results presented in a way that makes the data usable and transparent?

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the level of detail allow for an informed interpretation of the results?
• Are the data analyzed, compared, or presented in a way that is inappropriate or misleading?

BASIC CONSIDERATIONS

Considerations for this domain are highly variable depending on the outcomes of interest and must be refined by assessment teams.

A judgment and rationale for this domain should be given for each endpoint/outcome or group of endpoints/outcomes investigated in the study.

Is there concern that different types of tumors were inappropriately combined in the analysis?

Analyses of benign and malignant tumors from the same tissue type should be reported both separately and combined; tumors of the same cellular origin, which may appear at different organ sites, should be combined (McConnell et al. 1986).

Is there concern that statistical analyses are inadequate or were not conducted for evaluating the results?

If statistical analyses were not conducted, were the results reported in sufficient detail for ad hoc analysis?

If statistical analyses are not reported, the study should at a minimum present incidence data for specific tumors, so that statistical tests (e.g. Fisher’s exact test for pairwise comparisons) can be conducted.

If there is evidence of a decreased survival effect, the studies should use adequate statistical methods, such as the poly-3 test (Bailer and Portier 1988), to control for decreased survival. 

---

Sensitivity8535

Endpoint sensitivity14030

Is there concern that the animal model (source, species, strain, age, sex) is not sensitive for detecting an effect?

PROMPTING QUESTIONS/Elaboration

For each study:

• Did the study use an animal model that is sensitive for detecting tumors (e.g., known background tumor rates for the tumor type, sensitive to effects from the exposure route)?
• Studies in both sexes are more informative because a single sex study may miss cancers that are sex specific. Is there concern that an effect might have been missed because of a single sex study?
• Are there concerns regarding the specificity and validity of the protocols?
• Are there concerns regarding the timing of the endpoint assessment?

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study uses a known sensitive animal model for detecting potential carcinogenic effects.

Critical concerns

The study uses a known resistant animal model and no carcinogenic effect was detected.

EXAMPLE RATING

Experiments - No/minor concerns- The study authors report that the same species and strains have been used (by the same laboratory) before and have proven as sensitive for cancer hazard identification. The bioassay included animals of both sexes and exposure started at an appropriate age of the animals.

Statistical power14031

DESCRIPTION/Signaling question

Is there concern that the statistical power is inadequate (number of animals per exposure/dose and control groups to detect a neoplastic effect if present?

PROMPTING QUESTIONS/Elaboration

For each study: For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Did the study have adequate statistical power to detect induced low tumor incidences, the survival to study termination (in a treatment group), and the rarity of the induced tumors?

NOTE: Group sizes with a reasonable number of animals will be included. Sufficient sample size depends on the background rate of the tumor of interest. In transgenic animals, group sized of 10-15 can be sufficient. For cancer bioassays of 18 or more months, a sample size of 50 laboratory animals is considered sufficient because for most cancers that appear spontaneously in these animals in that time-frame, this sample size gives at least a 50% power for detecting a doubling of the background cancer risk (Portier and Hoel, 1983; Portier, 1986). 

Sample size alone is not a reason to conclude an individual study is critically deficient.

RATING CONSIDERATIONS (RoC)

No/minor concerns 

The study uses adequate numbers of animals for most tumor types.

 Critical concerns 

The study has a very small number of animals per dose group and no carcinogenic effect was detected.

Exposure contrast14032

DESCRIPTION/Signaling question

Is there concern that the exposure contrast is too low?

PROMPTING QUESTIONS/Elaboration

·      Were the body/tissue external metrics (external e-field, magnetic field or incident power flux density measured or calculated at the exposed body/tissue in the approximate far-field of the field source, and the exposure is at least 10 times (power flux density or field strength) above background level.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The exposure contrast is sufficient.

 Critical concerns

No exposure contrast OR the exposure contrast is assumed borderline.

Is there concern that the concurrent control group is not adequate for evaluating effects across treatment groups?14033

ELABORATION

Concurrent controls are considered to be the most relevant comparison group for evaluating potential exposure-related tumor effects. However, for rare tumor outcomes historical controls may enhance study sensitivity. 

FOLLOW-UP QUESTION:

NOTE: For rare tumor outcomes, the additional use of historical controls of the same strain, sub-strain, the same laboratory and a pre-defined time window can be used jointly with concurrent controls to increase study informativeness (IARC, 2006; IARC, 2019). 

No/minor concerns 

For rare tumor outcomes, appropriate historical controls have been used.

Critical concerns

For rare tumor outcomes appropriate historical controls have not been used.

Study duration14034

DESCRIPTION

Signaling question

Is the total study duration (exposure period and observation period) adequate to detect a neoplastic effect, if present?

PROMPTING QUESTIONS/Elaboration

For each study:

• Were outcomes measured after an appropriate latency period?

NOTE: while rodent cancer studies, in general, need to be at least one year in duration, there are exceptions depending upon the animal model and study design. Carcinogenicity studies in transgenic animals, co-carcinogen studies, or initiation/ promotion studies may need less time than one year for tumor development.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study duration is close to the lifetime of the animals, or an appropriate time period for the specific animal cancer model.

Critical concerns

The study duration is shorter than the appropriate time period for the animal model (e.g., less than one year for non-transgenic rats or mice) and no neoplasms are observed.

---

Major comments not covered in RoB or Sensitivity8573

Major comments not covered in RoB or Sensitivity14100

---

Selection8528

Allocation (randomization): Was administered dose or exposure level adequately randomized?14019

Randomization requires that each human subject or animal had an equal chance of being assigned to any study group including controls (e.g., use of random number table or computer-generated randomization). This applies to a concurrent negative control group (i.e., a group for which exposure is to vehicle or sham exposure or un-treated / cage controls) which must be included in the study to address randomization as well as any positive control group that may be part of the study.

PROMPTING QUESTIONS

For each study:

• Did each animal or litter have an equal chance of being assigned to any experimental group (i.e., random allocation)?
• Is the allocation method described?
• Aside from randomization, were any steps taken to balance variables across experimental groups during allocation?

Allocation (blinding): Was allocation to study groups adequately concealed at the start of the study?14020

Allocation concealment requires that research personnel do not know which administered dose or exposure level is assigned at the start of a study.

Allocation concealment prior to assigning the exposure level or treatment group (along with randomization in question #1) helps to assure that treatment is not given selectively based on potential differences in human subjects or non-human experimental animals. Allocation concealment requires that research personnel allocating subjects or animals to treatment groups (including the control group) could not foresee which administered dose or exposure level is going to be assigned at the start of a study.

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the study report blinding or other methods/procedures for reducing observational bias?
• If not, did the study use a design or approach for which such procedures can be inferred?
• What is the expected impact of failure to implement (or report implementation) of these methods/procedures on results?

Note:

1. a question under performance bias addresses blinding of personnel to treatment during the study;
2. a question under detection bias addresses blinding of outcome assessors.

---

Performance8530

Were the research personnel blinded to the study group during the study?14024

Blinding requires that study scientists do not know which administered dose or exposure level the animal is being given (i.e., study group).

---

Detection8532

Can we be confident in the exposure characterization?14026

Confidence requires valid, reliable, and sensitive methods to measure exposure applied consistently across groups.

It is essential that these criteria are considered, and potentially refined, by assessment teams, as the specific variables of concern can vary by exposure (e.g., reverberation chamber versus exposure coming from a cell phone).

PROMPTING QUESTIONS

·      Were body/tissue internal exposure metrics measured or calculated for the particular conditions in the study (SAR values (W/kg), SA (J/kg) induced electric field in (V/m), internal magnetic field strength in H/m or equivalent unit, respectively?

Can we be confident that possible body temperature increases have been properly assessed in the animals?14083

Did the study provide data on the body temperature of the animals?

PROMPTING QUESTIONS

For each study:

• Did the authors take steps to measure or calculate body temperature in the animals (i.e. in a pilot study)?
• Is the exposure very low and body temperature increases will be negligible (to be a risk of bias)?
• Are there concerns about significant increases in body temperature?

NOTE: There are no studies reporting that increases of body temperature by 1°C lead to cancer (There is limited evidence in experimental animals for the carcinogenicity of very hot water at 65 C or above, IARC vol 116), whereas other biological reactions and certain biomarkers can be changed due to temperature changes of 1°C (but none of these have been established as mechanisms of carcinogenicity).

Can we be confident in the outcome assessment?14027

Confidence requires valid, reliable, and sensitive methods to assess the outcome and the methods should be applied consistently across groups.

Detection bias can be minimized by using valid and reliable methods to assess the outcome applied consistently across groups (i.e., under the same method and time-frame). The objectivity of procedures used for measuring and reporting an outcome will impact the degree to which outcome assessors could bias the reported results.

NOTE: blinding of pathologists is not required nor common in academia, government agencies, industry and is not be considered as risk of bias. The initial review of slides is conducted unblinded (not masked), the secondary review is blinded (or masked) (Sills et al., 2019). This procedure is consistent with the position of the Society of Toxicologic Pathologists (STP) and the American College of Veterinary Pathologists (ACVP) (Crissman et al., 2004).

---

Attrition8531

Were outcome data complete without attrition or exclusion from analysis?14025

Attrition rates are required to be similar and uniformly low across groups with respect to withdrawal or exclusion from analysis.

Attrition or exclusion bias refers to systematic differences in the loss or exclusion from analyses of participants or animals from the study and how they were accounted for in the results.

PROMPTING QUESTIONS

For each study:

• Are all animals accounted for in the results?
• If there are discrepancies, do authors provide an explanation (e.g. death or unscheduled sacrifice during the study)?

If unexplained results omissions and/or attrition are identified, what is the expected impact on the interpretation of the results?

---

Selective Reporting8533

Were all measured outcomes reported?14028

Selective reporting is present if pre-specified outcomes are not reported or incompletely reported.

The reporting of the outcomes in this metric does not consider the quality of the data presentation, e.g., tranparency, description of the data (SEM or raw data).

Are the results and analysis presented in a in a way that makes the data usable and transparent?14084

CORE QUESTION

Are the results presented in a way that makes the data usable and transparent?

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the level of detail allow for an informed interpretation of the results?
• Are the data analyzed, compared, or presented in a way that is inappropriate or misleading?

BASIC CONSIDERATIONS

Considerations for this domain are highly variable depending on the outcomes of interest and must be refined by assessment teams.

A judgment and rationale for this domain should be given for each endpoint/outcome or group of endpoints/outcomes investigated in the study.

Is there concern that different types of tumors were inappropriately combined in the analysis?

Analyses of benign and malignant tumors from the same tissue type should be reported both separately and combined; tumors of the same cellular origin, which may appear at different organ sites, should be combined (McConnell et al. 1986).

Is there concern that statistical analyses are inadequate or were not conducted for evaluating the results?

If statistical analyses were not conducted, were the results reported in sufficient detail for ad hoc analysis?

If statistical analyses are not reported, the study should at a minimum present incidence data for specific tumors, so that statistical tests (e.g. Fisher’s exact test for pairwise comparisons) can be conducted.

If there is evidence of a decreased survival effect, the studies should use adequate statistical methods, such as the poly-3 test (Bailer and Portier 1988), to control for decreased survival. 

---

Sensitivity8535

Endpoint sensitivity14030

Is there concern that the animal model (source, species, strain, age, sex) is not sensitive for detecting an effect?

PROMPTING QUESTIONS/Elaboration

For each study:

• Did the study use an animal model that is sensitive for detecting tumors (e.g., known background tumor rates for the tumor type, sensitive to effects from the exposure route)?
• Studies in both sexes are more informative because a single sex study may miss cancers that are sex specific. Is there concern that an effect might have been missed because of a single sex study?
• Are there concerns regarding the specificity and validity of the protocols?
• Are there concerns regarding the timing of the endpoint assessment?

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study uses a known sensitive animal model for detecting potential carcinogenic effects.

Critical concerns

The study uses a known resistant animal model and no carcinogenic effect was detected.

EXAMPLE RATING

Experiments - No/minor concerns- The study authors report that the same species and strains have been used (by the same laboratory) before and have proven as sensitive for cancer hazard identification. The bioassay included animals of both sexes and exposure started at an appropriate age of the animals.

Statistical power14031

DESCRIPTION/Signaling question

Is there concern that the statistical power is inadequate (number of animals per exposure/dose and control groups to detect a neoplastic effect if present?

PROMPTING QUESTIONS/Elaboration

For each study: For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Did the study have adequate statistical power to detect induced low tumor incidences, the survival to study termination (in a treatment group), and the rarity of the induced tumors?

NOTE: Group sizes with a reasonable number of animals will be included. Sufficient sample size depends on the background rate of the tumor of interest. In transgenic animals, group sized of 10-15 can be sufficient. For cancer bioassays of 18 or more months, a sample size of 50 laboratory animals is considered sufficient because for most cancers that appear spontaneously in these animals in that time-frame, this sample size gives at least a 50% power for detecting a doubling of the background cancer risk (Portier and Hoel, 1983; Portier, 1986). 

Sample size alone is not a reason to conclude an individual study is critically deficient.

RATING CONSIDERATIONS (RoC)

No/minor concerns 

The study uses adequate numbers of animals for most tumor types.

 Critical concerns 

The study has a very small number of animals per dose group and no carcinogenic effect was detected.

Exposure contrast14032

DESCRIPTION/Signaling question

Is there concern that the exposure contrast is too low?

PROMPTING QUESTIONS/Elaboration

·      Were the body/tissue external metrics (external e-field, magnetic field or incident power flux density measured or calculated at the exposed body/tissue in the approximate far-field of the field source, and the exposure is at least 10 times (power flux density or field strength) above background level.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The exposure contrast is sufficient.

 Critical concerns

No exposure contrast OR the exposure contrast is assumed borderline.

Is there concern that the concurrent control group is not adequate for evaluating effects across treatment groups?14033

ELABORATION

Concurrent controls are considered to be the most relevant comparison group for evaluating potential exposure-related tumor effects. However, for rare tumor outcomes historical controls may enhance study sensitivity. 

FOLLOW-UP QUESTION:

NOTE: For rare tumor outcomes, the additional use of historical controls of the same strain, sub-strain, the same laboratory and a pre-defined time window can be used jointly with concurrent controls to increase study informativeness (IARC, 2006; IARC, 2019). 

No/minor concerns 

For rare tumor outcomes, appropriate historical controls have been used.

Critical concerns

For rare tumor outcomes appropriate historical controls have not been used.

Study duration14034

DESCRIPTION

Signaling question

Is the total study duration (exposure period and observation period) adequate to detect a neoplastic effect, if present?

PROMPTING QUESTIONS/Elaboration

For each study:

• Were outcomes measured after an appropriate latency period?

NOTE: while rodent cancer studies, in general, need to be at least one year in duration, there are exceptions depending upon the animal model and study design. Carcinogenicity studies in transgenic animals, co-carcinogen studies, or initiation/ promotion studies may need less time than one year for tumor development.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study duration is close to the lifetime of the animals, or an appropriate time period for the specific animal cancer model.

Critical concerns

The study duration is shorter than the appropriate time period for the animal model (e.g., less than one year for non-transgenic rats or mice) and no neoplasms are observed.

---

Major comments not covered in RoB or Sensitivity8573

Major comments not covered in RoB or Sensitivity14100

---

Selection8528

Allocation (randomization): Was administered dose or exposure level adequately randomized?14019

Randomization requires that each human subject or animal had an equal chance of being assigned to any study group including controls (e.g., use of random number table or computer-generated randomization). This applies to a concurrent negative control group (i.e., a group for which exposure is to vehicle or sham exposure or un-treated / cage controls) which must be included in the study to address randomization as well as any positive control group that may be part of the study.

PROMPTING QUESTIONS

For each study:

• Did each animal or litter have an equal chance of being assigned to any experimental group (i.e., random allocation)?
• Is the allocation method described?
• Aside from randomization, were any steps taken to balance variables across experimental groups during allocation?

Allocation (blinding): Was allocation to study groups adequately concealed at the start of the study?14020

Allocation concealment requires that research personnel do not know which administered dose or exposure level is assigned at the start of a study.

Allocation concealment prior to assigning the exposure level or treatment group (along with randomization in question #1) helps to assure that treatment is not given selectively based on potential differences in human subjects or non-human experimental animals. Allocation concealment requires that research personnel allocating subjects or animals to treatment groups (including the control group) could not foresee which administered dose or exposure level is going to be assigned at the start of a study.

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the study report blinding or other methods/procedures for reducing observational bias?
• If not, did the study use a design or approach for which such procedures can be inferred?
• What is the expected impact of failure to implement (or report implementation) of these methods/procedures on results?

Note:

1. a question under performance bias addresses blinding of personnel to treatment during the study;
2. a question under detection bias addresses blinding of outcome assessors.

---

Performance8530

Were experimental conditions identical across study groups?14023

Housing conditions and husbandry practices should be identical across control and experimental groups because these variables may impact the outcome of interest. Identical conditions include use of the same vehicle in control and experimental animals and the rating for this risk-of-bias element will depend largely on the consistent use vehicle across treatment groups. That is because under current reporting practices it is unlikely that similarity of conditions will be explicitly reported in most animal studies. Thus, we will assume unless stated otherwise that experimental conditions (other than treatment vehicle) were identical across groups.

Were the research personnel blinded to the study group during the study?14024

Blinding requires that study scientists do not know which administered dose or exposure level the animal is being given (i.e., study group).

---

Detection8532

Can we be confident in the exposure characterization?14026

Confidence requires valid, reliable, and sensitive methods to measure exposure applied consistently across groups.

It is essential that these criteria are considered, and potentially refined, by assessment teams, as the specific variables of concern can vary by exposure (e.g., reverberation chamber versus exposure coming from a cell phone).

PROMPTING QUESTIONS

·      Were body/tissue internal exposure metrics measured or calculated for the particular conditions in the study (SAR values (W/kg), SA (J/kg) induced electric field in (V/m), internal magnetic field strength in H/m or equivalent unit, respectively?

Can we be confident that possible body temperature increases have been properly assessed in the animals?14083

Did the study provide data on the body temperature of the animals?

PROMPTING QUESTIONS

For each study:

• Did the authors take steps to measure or calculate body temperature in the animals (i.e. in a pilot study)?
• Is the exposure very low and body temperature increases will be negligible (to be a risk of bias)?
• Are there concerns about significant increases in body temperature?

NOTE: There are no studies reporting that increases of body temperature by 1°C lead to cancer (There is limited evidence in experimental animals for the carcinogenicity of very hot water at 65 C or above, IARC vol 116), whereas other biological reactions and certain biomarkers can be changed due to temperature changes of 1°C (but none of these have been established as mechanisms of carcinogenicity).

Can we be confident in the outcome assessment?14027

Confidence requires valid, reliable, and sensitive methods to assess the outcome and the methods should be applied consistently across groups.

Detection bias can be minimized by using valid and reliable methods to assess the outcome applied consistently across groups (i.e., under the same method and time-frame). The objectivity of procedures used for measuring and reporting an outcome will impact the degree to which outcome assessors could bias the reported results.

NOTE: blinding of pathologists is not required nor common in academia, government agencies, industry and is not be considered as risk of bias. The initial review of slides is conducted unblinded (not masked), the secondary review is blinded (or masked) (Sills et al., 2019). This procedure is consistent with the position of the Society of Toxicologic Pathologists (STP) and the American College of Veterinary Pathologists (ACVP) (Crissman et al., 2004).

---

Attrition8531

Were outcome data complete without attrition or exclusion from analysis?14025

Attrition rates are required to be similar and uniformly low across groups with respect to withdrawal or exclusion from analysis.

Attrition or exclusion bias refers to systematic differences in the loss or exclusion from analyses of participants or animals from the study and how they were accounted for in the results.

PROMPTING QUESTIONS

For each study:

• Are all animals accounted for in the results?
• If there are discrepancies, do authors provide an explanation (e.g. death or unscheduled sacrifice during the study)?

If unexplained results omissions and/or attrition are identified, what is the expected impact on the interpretation of the results?

---

Selective Reporting8533

Were all measured outcomes reported?14028

Selective reporting is present if pre-specified outcomes are not reported or incompletely reported.

The reporting of the outcomes in this metric does not consider the quality of the data presentation, e.g., tranparency, description of the data (SEM or raw data).

Are the results and analysis presented in a in a way that makes the data usable and transparent?14084

CORE QUESTION

Are the results presented in a way that makes the data usable and transparent?

PROMPTING QUESTIONS

For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Does the level of detail allow for an informed interpretation of the results?
• Are the data analyzed, compared, or presented in a way that is inappropriate or misleading?

BASIC CONSIDERATIONS

Considerations for this domain are highly variable depending on the outcomes of interest and must be refined by assessment teams.

A judgment and rationale for this domain should be given for each endpoint/outcome or group of endpoints/outcomes investigated in the study.

Is there concern that different types of tumors were inappropriately combined in the analysis?

Analyses of benign and malignant tumors from the same tissue type should be reported both separately and combined; tumors of the same cellular origin, which may appear at different organ sites, should be combined (McConnell et al. 1986).

Is there concern that statistical analyses are inadequate or were not conducted for evaluating the results?

If statistical analyses were not conducted, were the results reported in sufficient detail for ad hoc analysis?

If statistical analyses are not reported, the study should at a minimum present incidence data for specific tumors, so that statistical tests (e.g. Fisher’s exact test for pairwise comparisons) can be conducted.

If there is evidence of a decreased survival effect, the studies should use adequate statistical methods, such as the poly-3 test (Bailer and Portier 1988), to control for decreased survival. 

---

Sensitivity8535

Endpoint sensitivity14030

Is there concern that the animal model (source, species, strain, age, sex) is not sensitive for detecting an effect?

PROMPTING QUESTIONS/Elaboration

For each study:

• Did the study use an animal model that is sensitive for detecting tumors (e.g., known background tumor rates for the tumor type, sensitive to effects from the exposure route)?
• Studies in both sexes are more informative because a single sex study may miss cancers that are sex specific. Is there concern that an effect might have been missed because of a single sex study?
• Are there concerns regarding the specificity and validity of the protocols?
• Are there concerns regarding the timing of the endpoint assessment?

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study uses a known sensitive animal model for detecting potential carcinogenic effects.

Critical concerns

The study uses a known resistant animal model and no carcinogenic effect was detected.

EXAMPLE RATING

Experiments - No/minor concerns- The study authors report that the same species and strains have been used (by the same laboratory) before and have proven as sensitive for cancer hazard identification. The bioassay included animals of both sexes and exposure started at an appropriate age of the animals.

Statistical power14031

DESCRIPTION/Signaling question

Is there concern that the statistical power is inadequate (number of animals per exposure/dose and control groups to detect a neoplastic effect if present?

PROMPTING QUESTIONS/Elaboration

For each study: For each endpoint/outcome or grouping of endpoints/outcomes in a study:

• Did the study have adequate statistical power to detect induced low tumor incidences, the survival to study termination (in a treatment group), and the rarity of the induced tumors?

NOTE: Group sizes with a reasonable number of animals will be included. Sufficient sample size depends on the background rate of the tumor of interest. In transgenic animals, group sized of 10-15 can be sufficient. For cancer bioassays of 18 or more months, a sample size of 50 laboratory animals is considered sufficient because for most cancers that appear spontaneously in these animals in that time-frame, this sample size gives at least a 50% power for detecting a doubling of the background cancer risk (Portier and Hoel, 1983; Portier, 1986). 

Sample size alone is not a reason to conclude an individual study is critically deficient.

RATING CONSIDERATIONS (RoC)

No/minor concerns 

The study uses adequate numbers of animals for most tumor types.

 Critical concerns 

The study has a very small number of animals per dose group and no carcinogenic effect was detected.

Exposure contrast14032

DESCRIPTION/Signaling question

Is there concern that the exposure contrast is too low?

PROMPTING QUESTIONS/Elaboration

·      Were the body/tissue external metrics (external e-field, magnetic field or incident power flux density measured or calculated at the exposed body/tissue in the approximate far-field of the field source, and the exposure is at least 10 times (power flux density or field strength) above background level.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The exposure contrast is sufficient.

 Critical concerns

No exposure contrast OR the exposure contrast is assumed borderline.

Is there concern that the concurrent control group is not adequate for evaluating effects across treatment groups?14033

ELABORATION

Concurrent controls are considered to be the most relevant comparison group for evaluating potential exposure-related tumor effects. However, for rare tumor outcomes historical controls may enhance study sensitivity. 

FOLLOW-UP QUESTION:

NOTE: For rare tumor outcomes, the additional use of historical controls of the same strain, sub-strain, the same laboratory and a pre-defined time window can be used jointly with concurrent controls to increase study informativeness (IARC, 2006; IARC, 2019). 

No/minor concerns 

For rare tumor outcomes, appropriate historical controls have been used.

Critical concerns

For rare tumor outcomes appropriate historical controls have not been used.

Study duration14034

DESCRIPTION

Signaling question

Is the total study duration (exposure period and observation period) adequate to detect a neoplastic effect, if present?

PROMPTING QUESTIONS/Elaboration

For each study:

• Were outcomes measured after an appropriate latency period?

NOTE: while rodent cancer studies, in general, need to be at least one year in duration, there are exceptions depending upon the animal model and study design. Carcinogenicity studies in transgenic animals, co-carcinogen studies, or initiation/ promotion studies may need less time than one year for tumor development.

RATING CONSIDERATIONS (RoC)

No/minor concerns

The study duration is close to the lifetime of the animals, or an appropriate time period for the specific animal cancer model.

Critical concerns

The study duration is shorter than the appropriate time period for the animal model (e.g., less than one year for non-transgenic rats or mice) and no neoplasms are observed.

---

Major comments not covered in RoB or Sensitivity8573

Major comments not covered in RoB or Sensitivity14100

---