How the "Conclusion" Transpired on Thyroid Cancer in the Second Round in Fukushima

Note: In translation of the official documents, "detection rate(s)" is used instead of "incidence rate" or "prevalence rate" in accordance with the original Japanese terminology.

Leaked "conclusion"

     On May 31, 2019, a Kyodo News article titled, "A Link Denied Between the Nuclear Power Plant Accident and Cancer in the Full-Scale Thyroid Screening in Children," appeared on the websites of Tokyo Shimbun and Okinawa Times. Below is an unofficial translation.
     An interview with an official on May 31 revealed that the expert subcommittee summarized an interim report reporting no link between cancer detected during the second round of the Thyroid Ultrasound Examination (TUE) and radiation exposure. The TUE covers all children in Fukushima Prefecture who were aged ≤ 18 years when the Tokyo Electric Power Company Fukushima Daiichi Nuclear Power Plant accident occurred, and the second round screening was conducted during FY 2014–2015. The reason for reporting no link is because there was no correlation showing an increased cancer rate at a higher exposure dose.
     As opposed to the first round that was intended to collect the baseline data, the second round is considered "Full-Scale Screening" which investigates effects of the accident on cancer. Now that opinions on the second round have been summarized for the first time, the future of the TUE might be impacted.

     This report came as a surprise to many of us who have been following the development of the Oversight Committee for the Fukushima Health Management Survey (FHMS) and one of its subcommittees, the Thyroid Examination Evaluation Subcommittee (herein, the TUE Subcommittee), for a simple reason that the analyses and deliberations conducted so far did not seem adequate to draw such a hasty conclusion. Discussions at the Subcommittee sessions have mostly been dominated by benefits and harm of the TUE itself, leading to an ongoing revision of the informed consent form. 
     It is true that an "interim report" was expected, as customary with any government committee, with the two-year term for the current TUE Subcommittee roster coming to a close as of the 13th TUE Subcommittee meeting held on June 3, 2019. What was not expected was that it would be such a hasty conclusion with a decisive tone.

"Draft" summary

     Chairperson Gen Suzuki started the meeting off complaining that the "scoop" headlines would take on a life of its own without conveying the process of discussion that led to the conclusion. The actual "interim report" released at the 13th Subcommittee meeting turned out to be only a "draft summary," and the conclusion was considered provisional.
     Accuracy of media reports released after the meeting seemed to depend on how well the writers have been informed of this complicated issue. Unfortunately, the only English article by Mainichi, "Thyroid cancer diagnoses in Fukushima youth not linked to nuke disaster: panel" did not accurately convey the original Japanese title "Thyroid cancer in Fukushima youth and radiation exposure 'not linked at this time': subcommittee" that emphasized the temporary nature of the conclusion. (Also the phrase in the Mainichi English article, "there is no data on those who have yet to be examined," is mistranslation of the original Japanese text that refers to lack of data on those who were examined outside the TUE system, explained as missing data in the previous post.

     Granted, the draft summary presented is only a draft. But the conclusion seems too decisive despite "provisional," considering the quality of analyses and discussions conducted so far. Some of the subcommittee members were not happy about the decisive tone of the draft summary, making suggestions to tone it down.
     In honor of Suzuki's wish to "convey the process of discussion that led to the conclusion," relevant documents have been translated into English. Translation of the draft summary is posted here, and translation of FMU's analyses used to draw the "provisional conclusion" will follow in the latter part of this post. 

     First, some background information is offered to characterize this TUE Subcommittee and "convey the process of discussion."

Analyses dominated by FMU
    
    When the TUE Subcommittee was reconvened after nearly a three-year hiatus in order to analyze the second-round data, there was an expectation that the analysis would be conducted by the subcommittee members which included experts in thyroid surgery, pediatrics, pathology, cancer statistics, epidemiology, and thyroid cancer research. However, the Subcommittee has been able to do little more than listening to Fukushima Medical University (FMU) officials presenting their own analyses of the data, mostly consisting of proportions without actual numerical results. 
    The subcommittee members have had to repeatedly ask for the actual data to no avail. When FMU officials said that after confounding factors were adjusted some results of analyses yielded a very small number, 1 or 2 cases, from very small municipalities that might make them identifiable, the members even offered, more than once, to hold a closed session to secure privacy of data. When FMU officials shared their hesitation to share raw data because it should not be looked at without removing confounding factors, chairperson Suzuki suggested that FMU share data at least with experts in statistics and epidemiology, if not the entire subcommittee, to help them sort through issues. 
     None of the advices or suggestions were taken up by FMU. Instead, they would submit analyses after analyses with their own interpretations. The subcommittee members would listen, ask question, offer some opinions and advises. Never once were the subcommittee members involved in analyzing the actual data simply because of the lack of opportunity. 

     (Note: During an informal interview immediately after the 13th Subcommittee press conference, chairperson Suzuki allegedly said that FMU withheld the actual raw data so as not to let a "certain magazine" publish its own analysis which would then take on a life of its own. Most certainly that magazine is Kagaku by Iwanami Publishers.)

Other futile discussions
     
     For some reason, when the TUE Subcommittee resumed, its members were still debating benefits and harm of the TUE itself even when the third round was fully under way. Recognition of the TUE as typical cancer screening, rather than health monitoring as described in the previous post, set the tone for the never-ending debate throughout the term of this TUE Subcommittee. The debate even intensified after the IARC expert group TM-NUC (Thyroid Monitoring after Nuclear Accident) released its reportrecommending against systematic thyroid screening for fear of overdiagnosis. (See this 2017 post or previous post for more on TM-NUC.) 
     Further, Toru Takano, a thyroid cancer researcher from Osaka University and the only subcommittee member doubly appointed to the Oversight Committee, kept insisting that the TUE administered in school-settings (in elementary and junior high schools) constituted an ethical and human rights violation because students were "forced" into it. (The truth is that the school-based TUE, approved by FMU's ethics committee, had been requested by municipal education commissions to prevent missed school days and extra load on families from taking children to a specified TUE location. Only students with signed consent forms are examined.) 
     According to Takano's own hypothesis of fetal cell carcinogenesis, all children have "self-limiting thyroid cancer." Hence all thyroid cancer cases detected in Fukushima comprise overdiagnosis, so the TUE should immediately be stopped and replaced by palpation of the neck. Why Japan Thyroid Association appointed Takano to the double positions is unclear. If the intention was to create disturbance and muddle the debate, their choice certainly nailed it.

Disappearing regional differences

     As described in the previous post, the TUE data has numerous transparency and reliability issues. Still, some meaningful analysis on the official data could be of use. 
     The final report of the second round has never been translated into English because it was first presented to the 8th TUE Subcommittee as Document 2-1. (Documents from subcommittee are never officially translated.) Included in the final report is Table 11 for a regional analysis which, unlike in the first round (see Table 9), appears to show regional differences in detection rates of thyroid cancer in the second round (shown in the bottom rows). (See the comparison between Tables 11 and 9 in this post.)
     Also presented at the 8th TUE Subcommittee meeting as Document 2-3 was another set of the regional analysis adjusted with screening interval, the length of time between primary examinations of the first and second rounds. This analysis only includes 246,687 subjects who participated in both of the first and second rounds, eliminating 23,829 or nearly 9% of 270,516 participants. 
     Table 11 and the version adjusted for screening interval is shown below.

The second-round results by region: original Table 11 and adjusted for screening interval by on Scribd

  
    Other than this adjustment for screening interval, FMU has compared the first- and second-round data (see Document 2 from the 10th Subcommittee) and done just about everything seemingly possible to "adjust" detection rates of confirmed or suspected cancer for various factors which can affect the second-round data (see Document 1 from the 11th Subcommittee). Excerpts fro these 2 documents were presented as Document 1-1 at the 13th Subcommittee meeting.

Dose-response analysis with UNSCEAR doses

     Their effort culminated in a dose-response analysis using absorbed doses to the thyroid by municipality estimated by UNSCEAR. Presented as Document 1-2 at the 12th Subcommittee meeting on February 22, 2019, it claimed no dose-response in confirmed or suspected cancer cases from the second round. Of course, the best doses to be used are individual thyroid doses, but only 1080 children from 3 municipalities had their thyroids directly measured (and these measurements are likely underestimated). UNSCEAR estimated doses by municipality were considered by chairperson Suzuki the only "authorized" doses available at present to be used in a dose-response analysis, despite reservations about its use as expressed by experts from National Cancer Center Japan.
     This dose-response analysis has been severely criticized by Junichiro Makino, an astromist/physicist at Kobe University, for questionable methods in a Japanese science journal Kagaku (only available in Japanese). Makino has previously pointed out in his 2015 book "Assessment of radiation exposure and scientific methods" that thyroid absorption doses estimated by UNSCEAR have a significant amount of uncertainties because they are essentially based on deposition of radioactive cesium 134/137. 
     It is not radioactive cesium but radioactive iodine that affects the thyroid gland. However, due to a very short half-life (8 days) of radioactive iodine 131 and even shorter half-life (2.3 hours) of iodine 132 existing in equilibrium with tellurium 132 (half-life 3.2 days), it is extremely difficult to conduct actual measurements of radiological contamination due to short-lived iodine isotopes. Thus UNSCEAR-estimated thyroid absorption doses were calculated from radioactive cesium levels based on a theoretically-derived cesium-to-iodine ratio which in reality may not remain the same under different circumstances.
     In short, results from the dose-response analysis using UNSCEAR estimated doses would not be considered very reliable due to a significant amount of inherent uncertainties

Underestimation due to an error 

     At the June 3, 2019 TUE Subcommittee meeting, FMU admitted that an error made while entering data in analytical program of statistical software had produced odds ratios which were smaller than actual values. correction (Document 1-3, translation shown below) was issued that showed bigger odds ratios and wider 95% confidence intervals. Insisting that the error does not change the results of significant testing, FMU upholds their conclusion of no dose-response. 
     As to preventing future errors, FMU formed an investigative committee to look into the matter and decided to establish a system where two individuals would conduct the same analysis and compare results. With their competency in question, a better option might be a third-party analysis and oversight.

New analysis 

    As if to back up their "no dose-response" claim, FMU also released further odds ratio analyses (Document 1-2, translation shown below) with an adjustment of various factors. The "new" analysis released at the same time as the draft summary was bizarre. 
     The subcommittee members appeared literally at a loss. In fact, some stated that it was difficult to interpret the results without seeing the number of cases in each dose group. Others seemed to be having difficulty digesting FMU's own analyses and interpretation in a short time. 
     In particular, some of the graphs actually show a negative trend, with odds ratios falling below 1. Testuya Ohira, an FMU official, explains, "a negative trend seen is an unlikely item in general, so for now we are calling it no dose-response."

⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨⬨

     At the 12th Subcommittee meeting in February 2019, it was becoming clear that the TUE Subcommittee was running out of time to make a proper assessment before the two-year term was up, but in our naivety, some of us anticipated that discussion might be carried onto the next term without drawing any definitive conclusion. 

     However, what actually took place at the very last meeting of the current term was beyond words: Based on FMU's analysis criticized for lack of transparency or comprehensibility from start to finish, chairperson Suzuki put together a draft summary with a provisional conclusion of no radiation effects for thyroid cancer detected in the second round
     When asked to "grade" the conclusion at the press conference, Tomotaka Sobue, a cancer epidemiologist, gave 60 out of 100, and Kota Katanoda, a cancer statistician, gave 50. They both pointed to 2 things that account for the tentative nature of the conclusion, 1) lack of individual doses and 2) missing data on cancer cases diagnosed during the clinical follow-up or outside the TUE.  
     Suzuki says the provisional conclusion is to be treated with caution because the second-round data will eventually be re-analyzed. Unfortunately, once it hits a headline, the provisional conclusion is likely to skip whatever nuance is intended and take on a life of its own. 

Translation of the FMU analyses

     Below is the translation of 3 documents presented to the 13th TUE Subcommittee on June 3, 2019. Showing analyses by FMU, these documents comprise the basis for the "provisional conclusion" that denies radiation effects for the second round of the TUE. 



No comments:

How the "Conclusion" Transpired on Thyroid Cancer in the Second Round in Fukushima

Note: In translation of the official documents, "detection rate(s)" is used instead of "incidence rate" or "prevale...