NTP Lite? How Lite Is It Really?
“NTP Lite” is an informal shorthand used to refer to two new animal studies from Japan and South Korea, published in 2026. Both studies reported no statistically significant increase in tumour incidence following long-term exposure to 900 MHz radiofrequency electromagnetic fields (RF-EMF).[1,2]
Their findings are particularly relevant because an international re-evaluation of RF-EMF by the International Agency for Research on Cancer (IARC) has been recommended as a high priority for the 2025–2029 period.[3] Evidence from experimental animal studies will form part of that assessment.
However, assessments by scientific and regulatory bodies in Switzerland, Germany and Australia have highlighted different aspects of the findings. While the studies have been welcomed as relevant new evidence, questions have also been raised about their design, statistical power and ability to investigate the tumour findings reported by the US National Toxicology Program (NTP) in 2018.
How much can these new studies tell us? And are they sufficiently robust to investigate the findings that prompted the NTP’s conclusion of clear evidence of carcinogenic activity in male rats?[4]
How Have Scientific and Regulatory Bodies Assessed the 2026 Studies?
BERENIS – Swiss Expert Group on Electromagnetic Fields and Non-Ionising Radiation
In its Newsletter No. 46 (October 2026), BERENIS assessed the two 2026 studies from Japan (Imaida et al.) and South Korea (Kim HS et al.) in relation to the 2018 NTP study.
Its assessment focused on three main issues:
Differences in exposure levels and study design
BERENIS highlighted important differences between the exposure levels and study designs of the 2026 studies and the NTP research. These are examined in more detail in the comparison section below.
Survival differences
BERENIS raised concerns about the Japanese study’s interpretation of tumour findings in relation to differences in survival between the exposed and sham-exposed animals. We return to this issue when examining the survival data.
Statistical power and rare tumours
BERENIS questioned whether the relatively small groups of animals in the 2026 studies provided sufficient statistical power to investigate the rare tumours identified in the NTP research. This issue is considered in more detail in the section on statistical power.
BERENIS’s overall conclusion
BERENIS questioned whether the two studies could properly be regarded as replications of the NTP research. Its conclusion was explicit:
“These studies never had the potential to verify the results of the NTP study, which, according to the authors, was one of the main aims.”
It also noted that the publications made no mention of consultation with NTP researchers regarding study design, exposure or dosimetry.[5]
BfS – Germany
The German Federal Office for Radiation Protection (BfS) described the Japan–South Korea collaboration as a valuable approach, but cautioned that uncertainties remain.
Neither study found a statistically significant increase in tumour incidence. However, when the results of the two studies were combined, the RF-exposed groups had three cardiac schwannomas and two brain gliomas, compared with one of each in the sham-exposed groups.
BfS noted that the tumour findings in the two studies pointed in the same direction. However, the wide 95% confidence intervals and limited statistical power meant that the findings did not necessarily contradict the NTP results.[6]
BfS also highlighted the differences in exposure levels and group sizes between the collaborative studies and the NTP research. It concluded that it remains uncertain whether statistically significant increases would have been detected had higher exposure levels and/or larger groups of animals been used.[6]
ARPANSA – Australia
The Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) published its assessment of the Japanese and South Korean studies in January 2026. The webpage was subsequently reviewed on 25 February 2026.
ARPANSA noted that the studies were conducted simultaneously but independently, using a common research protocol. Both followed Good Laboratory Practice and OECD guidelines. Neither study found a statistically significant increase in tumour incidence or evidence of DNA damage or chromosomal aberrations.
ARPANSA concluded that the findings did not substantiate the earlier NTP results. It also noted that the tumour histopathology had undergone international peer review. This does not mean that ARPANSA’s own assessment was peer reviewed.[7]
What Did the Original NTP Study Find?
In male rats exposed to CDMA-modulated RF radiation, the NTP reported two findings of particular relevance.
Malignant schwannomas of the heart
No malignant cardiac schwannomas occurred in the sham-control group. They were found in all three exposed groups:
- 1.5 W/kg: 2 of 90 rats
- 3 W/kg: 3 of 90 rats
- 6 W/kg: 6 of 90 rats.[4]
The increase at 6 W/kg was statistically significant, as was the positive exposure-related trend.[4]
Malignant gliomas of the brain
No malignant gliomas occurred in the sham-control group or the 1.5 and 3 W/kg groups. Three occurred in the 6 W/kg group. Although this increase was not statistically significant in the individual group comparison, the NTP considered the findings related to RF exposure.[4]
The NTP’s conclusion
The NTP concluded that there was clear evidence of carcinogenic activity in male rats exposed to CDMA-modulated RF radiation.
- Cardiac schwannomas formed the basis for this conclusion.
- Brain gliomas were also considered related to exposure, but did not form the basis for the clear evidence classification.[4]
How Do the 2026 Studies Compare with the NTP?
Both 2026 studies used male Hsd:Sprague Dawley SD rats. Exposure began on gestational day 5 in the pregnant dams and continued after birth. At weaning, male offspring were selected for the two-year carcinogenicity study and continued to receive exposure until approximately 104 weeks of age.[1,2]
The researchers used a 900 MHz CDMA-modulated RF signal at a whole-body specific absorption rate (SAR) of 4 W/kg. Exposure was delivered for 18 hours and 20 minutes per day in 10-minute on/off cycles.[1,2]
The NTP study included both male and female rats and investigated GSM and CDMA signals at three whole-body SAR levels: 1.5, 3 and 6 W/kg. The 2026 studies included only male rats, investigated CDMA signals only, and used a single exposure level of 4 W/kg.[1,2,4]
Using only one exposure level meant that the researchers could not examine a dose-response relationship. The 2026 studies also did not include the NTP’s highest exposure level of 6 W/kg, at which the statistically significant increase in malignant cardiac schwannomas was observed.[4]
The group sizes were also smaller. For the two-year carcinogenicity phase, the researchers assigned 70 male rats to each group in both 2026 studies, compared with 90 animals per group in the NTP study.[1,2,4]
These differences are important when considering whether the newer studies had sufficient statistical power to investigate the specific tumour findings reported by the NTP.
Interpreting the Tumour Findings
Controls and Survival
A control group is not simply a group in which nothing happens. It provides the baseline against which the exposed animals are assessed.
Spontaneous tumours
Rats can develop tumours without experimental exposure, particularly as they age. Some tumours, including cardiac schwannomas and brain gliomas, are rare.[1,4]
The presence of tumours in a control group does not automatically mean that an exposure has no effect. Researchers compare tumour incidence in the exposed and control groups and also consider historical control data. These data show the range of tumour incidences observed in comparable animals in previous studies. They provide useful context, particularly for rare tumours, but do not replace the concurrent control group.[1,2,4]
Sham controls and cage controls
The 2026 studies included both sham-exposed and cage-control animals.[1,2]
Sham-exposed animals underwent the same chamber procedures as the RF-exposed animals, but without active RF exposure. Cage-control animals remained under normal housing conditions and did not undergo the chamber exposure procedure.[1,2]
Including both groups allowed researchers to compare the effects of RF exposure with those of the experimental procedures and normal housing conditions.
Survival differences in the NTP study
The male sham-control rats in the NTP study had substantially lower survival than the RF-exposed groups. For CDMA exposure, the estimated two-year survival probability was approximately 28% in the sham-control group, compared with 48%, 62% and 48% in the 1.5, 3 and 6 W/kg groups, respectively.[4]
The NTP attributed the lower survival in the sham-control group largely to more severe chronic progressive nephropathy, a progressive kidney disease commonly found in ageing rats.[4]
This matters because animals that survive longer have more opportunity to develop tumours, particularly those that occur later in life. The NTP therefore used statistical methods, including the Poly-3 test, to account for differences in survival when analysing tumour incidence.[4]
Survival differences in the 2026 studies
The Japanese study also reported a survival difference, but in the same direction as the NTP findings: more RF-exposed animals survived than sham controls.[1]
At 105 weeks, 44 of 68 RF-exposed rats (64.7%) survived, compared with 30 of 70 sham-exposed rats (42.9%). In the cage-control group, 24 of 70 animals (34.3%) survived.[1] The higher survival in the RF-exposed group was statistically significant.[1]
Two RF-exposed animals died at week 20 following equipment malfunction and were excluded from the survival analysis. This explains why the RF-exposed group had 68 animals available for the analysis, rather than 70.[1]
The Japanese researchers used Peto’s test to analyse tumour incidence.[1] They also discussed the lower body weight and food consumption observed in the RF-exposed animals as possible factors associated with the survival difference.
BERENIS criticised the Japanese authors’ use of the survival difference to dismiss tumour findings generally, regardless of the direction in which those findings pointed.[5]
The Korean study also included 70 animals initially in each of the three groups. At 104 weeks, survival was 58.6% in the RF-exposed and cage-control groups and 54.3% in the sham-exposed group. The researchers reported no statistically significant differences in survival between the three groups.[2]
Does No Statistical Significance Mean There Is No Effect?
A finding of no statistically significant increase in tumours does not necessarily mean that no increase occurred. It may also mean that a study did not have sufficient statistical power to detect an effect, particularly when the tumours being investigated are rare.
Why statistical power matters
Statistical power refers to a study’s ability to detect an effect if one exists. The smaller the expected effect and the rarer the tumour, the more difficult it becomes to detect a difference between exposed and control animals.
What does the 4.5% estimate tell us?
Based on the NTP findings, BERENIS estimated that the expected response for malignant cardiac schwannomas at 4 W/kg would be approximately 4.5%.[5]
This was an estimate derived from the NTP findings; it was not an observed result in either of the 2026 studies. It illustrates how small the expected response might be and why detecting it would be difficult with relatively small groups of animals.
What about a 20% increase?
BERENIS also drew attention to the Korean researchers’ observation that an increase of around 20% in the incidence of cardiac schwannomas or malignant gliomas would not be readily detectable with the sample size used.[5]
These considerations are important when assessing what the absence of statistically significant findings in the 2026 studies can tell us.
What Do the 2026 Studies Tell Us?
Taken together, the 2026 studies provide relevant new evidence, but they do not provide a definitive test of the NTP findings. Several points are important when interpreting their results:
- The studies are relevant and should not be dismissed. They add to the scientific evidence on RF-EMF exposure and carcinogenicity.
- They did not reproduce the full NTP exposure design. They used a single exposure level of 4 W/kg, rather than the three exposure levels used by the NTP, including its highest level of 6 W/kg.
- Their statistical power was limited. The smaller groups of animals make it difficult to detect increases in rare tumours.
- They do not settle the question raised by the NTP study. The absence of statistically significant increases does not establish that the NTP findings have been disproved or that a smaller increase in tumour incidence can be ruled out.
EMFSA | 7 November 2026
References
- Imaida K, et al. (2026). The International Collaborative Animal Study of mobile phone radiofrequency radiation carcinogenicity and genotoxicity: the Japanese study. Toxicological Sciences, 209(3), kfag002. https://doi.org/10.1093/toxsci/kfag002
- Kim HS, et al. (2026). The International Collaborative Animal Study of the carcinogenicity and genotoxicity of mobile phone radiofrequency radiation: the Korean study. Toxicological Sciences, 209(3), kfag001. https://doi.org/10.1093/toxsci/kfag001
- International Agency for Research on Cancer (IARC). (2024). Report of the Advisory Group to Recommend Priorities for the IARC Monographs during 2025–2029. Read report (PDF)
- National Toxicology Program (NTP). (2018). Toxicology and Carcinogenesis Studies in Hsd:Sprague Dawley SD Rats Exposed to Whole-Body Radio Frequency Radiation at a Frequency (900 MHz) and Modulations (GSM and CDMA) Used by Cell Phones. NTP Technical Report 595. Read report (PDF)
- BERENIS. (October 2026). Newsletter No. 46. Swiss Expert Group on Electromagnetic Fields and Non-Ionising Radiation. Read newsletter (PDF)
- Federal Office for Radiation Protection (BfS), Competence Centre for Electromagnetic Fields (KEMF). (2026). Spotlight on “The international collaborative animal study of the carcinogenicity and genotoxicity of mobile phone radiofrequency radiation” by Imaida et al. (Japanese study) and Kim et al. (Korean study) in Toxicological Sciences (2026). Spotlight – Jun/2026 No. 1 (Eng). URN: urn:nbn:de:0221-2026061660652.
- Australian Radiation Protection and Nuclear Safety Agency (ARPANSA). (January 2026; webpage reviewed 25 February 2026). Korean and Japanese studies agree that radiofrequency exposure does not cause cancer in rats. Read ARPANSA assessment