Neurology -- Correspondence for Horner et al., 61 (6) 742-749

Correspondence published:

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans: Carmel Armon (11 December 2006)

Reply from the Authors: Ronnie D. Horner, John R. Feussner, Edward J. Kasarskis, Steven C. Grambow (11 December 2006)

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans: Steven R. Brenner (27 October 2003)

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans: Carmel Armon, MD, MHS (27 October 2003)

Reply to Brenner and Armon: Ronnie Horner, PhD, Ronnie D. Horner, PhD and John R. Feussner, MD, MPH for the Investigators of VA Cooperative Studies Program Project #500 (27 October 2003)

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans 11 December 2006

Carmel Armon,
Division of Neurology, Baystate Medical Center
759 Chestnut Street, Springfield MA 01199
Send Correspondence to journal:
Re: Occurrence of amyotrophic lateral sclerosis among Gulf War veterans

carmel.armon{at}bhs.org Carmel Armon

I recently read an article [1] by authors affiliated with the United States Department of Veterans� Affairs and the United Kingdom Ministry of Defense. It provides data relevant to the first report of an excess of ALS cases in the deployed population. [2] The relevant text [1] is:
"The findings of the ALS study have had to be qualified by two considerations:
(i) the results could have been influenced by ascertainment bias because they were based on just seven additional cases of ALS among Gulf War veterans in a study of 2.5 million participants; and
(ii) the mortality rate of Gulf War veterans due to ALS has not yet been found to be elevated [3]."
I had not previously read or heard the statement that the entire claim of excess incidence of ALS in deployed veterans hinged on seven additional cases of ALS. This information was not shared with the readers of Neurology.
In addition, since publication of the first report [2], the original authors have confirmed that there was underascertainment in the non- deployed First Gulf War population. [4] Compared to 67 cases reported in the original article, the range of point estimates derived using three capture-recapture methods was 70-81 cases, However, when the authors used the Washington state incidence estimates to derive an upper bound for the number of cases expected in the non-deployed population � it was as high as 90 or 95. The higher numbers are not reflected in the abstract of the article [4], and were not used by the authors in recalculating their estimates of the ALS incidence ratio of deployed and non-deployed veterans. They used the lower estimates.
Thus, the authors� claim of excess cases in the deployed population continues to hinge on accepting as valid a lower-than-expected number in the non-deployed population (relative to a well-ascertained US population), and is supported by methods that side step the implications of what appears to be at most, a very small absolute number of excess cases.
This additional information appeared belatedly in publications most readers of Neurology may not have read. [1,4] Furthermore, it is not readily retrievable because it is not reflected in the abstracts of those publications. It supports my original critique. [5]
References
1. Hyams KC, Brown M, White DS. Resolving disputes about toxicological risks during military conflict: The US Gulf War Experience. Toxicology Reviews 2005;24:167-180.
2. Horner RD, Kamins KG, Feussner JR, et al. Occurrence of amyotrophic lateral sclerosis among Gulf War veterans. Neurology, 2003;61:742-749.
3. Rose MR. Gulf War service is an uncertain trigger for ALS. Neurology. 2003;61:730-731.
4. Coffman CJ, Horner RD, Grambow SC, Lindquist J; VA Cooperative Studies Program Project #500. Estimating the occurrence of amyotrophic lateral sclerosis among Gulf War (1990-1991) veterans using capture- recapture methods. Neuroepidemiology 2005;24:141-150.
5. Armon C. Occurrence of amyotrophic lateral sclerosis among Gulf War veterans (Correspondence). Neurology. 2004;62:1027-1028.
The author reports no conflicts of interest.

Reply from the Authors 11 December 2006

Ronnie D. Horner,
University of Cincinnati Institute for the Study of Health
French-East Bldg, Suite 275, 3202 Eden Ave., Cincinnati, OH 45267,
John R. Feussner, Edward J. Kasarskis, Steven C. Grambow
Send Correspondence to journal:
Re: Reply from the Authors

ronnie.horner{at}uc.edu Ronnie D. Horner, et al.

We welcome this opportunity to again respond to the two major points of contention regarding our finding of a two-fold increase in risk of ALS among 1991 Gulf War veterans. [2] These include under-ascertainment of cases among the non-deployed military personnel (i.e., the control group) and failure to use a general population as the control group.
We are aware that there may have been under-ascertainment of cases among the non-deployed military personnel who were on active duty during the 1991 Gulf War. Therefore, as Dr. Armon points out, we assessed the presence and magnitude of case under-ascertainment using capture-recapture methodology. [4] Details of the analysis can be found in that report.
Only when the number of cases among the non-deployed is projected to be 103 or greater (versus the 67 cases actually identified) does the 95% confidence limit on the rate ratio become statistically non- significant. We contend that this number of cases is unlikely as the rate of ALS in this group would exceed that of a general population, the report of military service being a risk factor for ALS not withstanding. [6]
Regarding the use of a general population as the control group, military service is an occupation and the first principles of occupational epidemiology indicate that the most appropriate comparison group for a study of ALS among deployed military personnel is another group of military personnel. This is because entry into military service involves a selective process. A general population has not undergone this selective process and does not qualify as an appropriate comparison group. Moreover, as far as we can determine, we have the largest series of ALS cases who are under 55 years of age. Our rates are likely to be the most stable rates available for this age range making our rate estimates more reliable than those from general populations.
Finally, the lack of an elevated risk of death from this rare disease among Gulf War veterans is a likely consequence of insufficient follow-up time; the most recent mortality study involves only 7 years of follow-up. [7]
We are gratified that while we academicians haggle about methodological nuances to a degree befitting Talmudic scholars, former VA Secretary Anthony Principi acted on our findings by changing VA policy to make Gulf War veterans with ALS eligible for VA benefits.
References

6. Weisskopf MG, O�Reilly EJ, McCullough ML et al. Prospective study of military service and mortality from ALS. Neurology 2005; 64:32-37.
7. Kang HK, Bullman TA. Mortality among US veterans of the Persian Gulf War: 7-year follow-up. Am J Epidemiol 2001; 54:399-405.
Disclosure: The authors disclose no conflicts of interest.

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans 27 October 2003

Steven R. Brenner,
Saint Louis VA Medical Center Neurology Department and Saint Louis University Neurology Department
Department Neurology, Saint Louis VA Medical Center, 915 North Grand, Saint Louis, MO 63106
Send Correspondence to journal:
Re: Occurrence of amyotrophic lateral sclerosis among Gulf War veterans

SBren20979{at}aol.com Steven R. Brenner

I read the article on the occurrence of amyotrophic lateral sclerosis (ALS) among Gulf War veterans with interest. [1] During part of the Gulf war period, an epidemic of aplastic crisis [2] occurred between July, 1991 and March, 1992 in the Gulf region. Forty-eight patients with aplastic crises were hospitalized at Saudi Aramco-Dhahrain Health Center, Saudi Arabia with evidence of recent human parvovirus (PV) B19 infection in 91% of the cases.
The epidemic raises the possibility of exposure to PV in the Gulf region during the Gulf war period, PV being a virus spread by respiratory means. It causes Fifth disease in children.
The autonomous PV minute virus of mice large nonstructural protein NS1 can directly interact in vitro and in vivo with human survival motor neuron gene (SMN). [3] An abnormal SMN1 gene locus may be a susceptibility factor for ALS [4], since the abnormal copy number of the gene was present in 16% of ALS patients and in only 4% of controls.
PV among Gulf War personnel may have led to interaction with the SMN gene and contributed to development of ALS.
Mechanisms by which viruses may interact with the host cell genome and with the human immune system make the viral hypothesis of ALS still warrant further studies. [5]
References
1. Horner RD, Kamins KG, Feussner SC, Grambow SC, et al. Occurrence of amyotrophic lateral sclerosis among Gulf War veterans. Neurology . 2003; 61: 742-749.
2. Mullouh AA, Qudah A. An epidemic of aplastic crisis caused by human parvovirus B19. Pediatr. Infect Dis J. 1995; 14: 31-4.
3. Young PJ, Jensen KT, Burger LR, Pintel DJ, Lorson CL. Minute virus of mice NS1 interacts with the SMN protein, and they colocalize in novel nuclear bodies induced by parvovirus infection. J Virol. 2002: 76:3892- 904.
4. Corcia P, Mayeux-Portas V, Hohoris J, et.al ; French ALS Research Group. Amyotrophic Lateral Sclerosis. Abnormal SMN1 gene copy number is a susceptibility factor for amyotrophic lateral sclerosis. Ann Neurol. 2002; 51:243-6.
5. Sola P, Bedin R, Casoni F, Barozzi P, et al. New insights into the viral theory of amyotrophic lateral sclerosis: study on the possible role of Kaposi�s sarcoma-associated virus/human herpesvirus 8. Eur Neurol. 2002; 47: 108-12.

Occurrence of amyotrophic lateral sclerosis among Gulf War veterans 27 October 2003

Carmel Armon, MD, MHS,
Baystate Medical Center
Springfield, MA
Send Correspondence to journal:
Re: Occurrence of amyotrophic lateral sclerosis among Gulf War veterans

carmel.armon{at}bhs.org Carmel Armon, MD, MHS

We read the article by Horner et al with interest. [1] Neurology readers now have data that permit evaluating the validity of the study�s conclusions.
The chief flaw in the study design was taking, as the primary outcome variable, the ratio between the incidences of ALS in the deployed and the non-deployed veterans. A lower-than-usual number of cases identified in the non-deployed veterans would make even a �usual� number of cases in the deployed veterans appear excessive.
The authors attempt to convince the readers that case finding was complete. For example, capture-recapture methodology was applied to estimate completeness of ascertainment; however, in my opinion, this is most informative when the data sources are relatively independent, which was not the case in the present study. In addition, the data were reanalyzed excluding the groups identified by passive ascertainment alone. This approach cannot compensate for not knowing how many patients were not identified.
Finally, average overall incidence rates standardized to the 1990 US population are presented. Reservations about the appropriateness and accuracy of the calculations notwithstanding, these numbers may be compared directly to those found in three population-based studies [2-4] (Table). The adjusted overall rates in the non-deployed veterans are lower by approximately 30% than those in the comparison groups. In my opinion, the most compelling evidence in rebuttal of the claim that ALS occurred excessively in 1990 Gulf Region veterans, particularly in those under 45 years old, in whom 89 of the 107 total cases arose, is direct comparison to age-and-gender-specific data from the western Washington state population-based ALS incidence study [2] (Table, bold type). The rates in the under 45-year-old Gulf War veterans� age groups are approximately half those of the Washington state under 45-year-old population. Even if the risk in the deployed population were truly double that of the non-deployed population, it would still be less than that of the same age groups in Washington state. Taken at their face value, these data might lead to the conclusion that some forms of military service in 1990 protected against the occurrence of ALS, rather than that some forms of military service placed individuals at increased risk of occurrence of ALS.
The study design also failed to take into consideration the possibility of confounding by smoking [5-6], and that the biologic onset of ALS precedes the clinical onset, probably by several years. [7-8] Consequently, the present data do not permit concluding that the incidence of ALS in veterans who were deployed to the Gulf Region in 1990 has been any greater than in an appropriately matched continental US population. The major flaw in the inception of the study was that the hypothesis driving it lacked biologic plausibility. I agree with Professor Rose, [9] that further research in this population is unlikely to yield etiologically useful data.
Table
References
1. Horner RD, Kamins KG, Feussner JR et al. Occurrence of amyotrophic lateral sclerosis in Gulf War veterans. Neurology 2003; 61:743-749.
2. McGuire V, Longstreth WT Jr, Koepsell TD, van Belle G. Incidence of amyotrophic lateral sclerosis in three counties in western Washington state. Neurology. 1996 ;47:571-3.
3. Sorenson EJ, Stalker AP, Kurland LT, Windebank AJ. Amyotrophic lateral sclerosis in Olmsted County, Minnesota, 1925 to 1998. Neurology. 2002 ;59:280-2.
4. Mandrioli J, Faglioni P, Merelli E, Sola P. The epidemiology of ALS in Modena, Italy. Neurology. 2003;60:683-9.
5. Kamel F, Umbach DM, Munsat TL, Shefner JM, Sandler DP. Association of cigarette smoking with amyotrophic lateral sclerosis. Neuroepidemiology. 1999;18:194-202.
6. Nelson LM, McGuire V, Longstreth WT Jr, Matkin C. Population-based case-control study of amyotrophic lateral sclerosis in western Washington State. I. Cigarette smoking and alcohol consumption. Am J Epidemiol. 2000 ;151:156-63.
7. Kurland LT, Radhakrishnan K, Smith GE, Armon C, Nemetz PN. Mechanical trauma as a risk factor in classic amyotrophic lateral sclerosis: lack of epidemiologic evidence. J Neurol Sci 1992;113:133-43.
8. Armon C. An Evidence-Based Medicine Approach to the Evaluation of the Role of Exogenous Risk Factors in Sporadic Amyotrophic Lateral Sclerosis. Neuroepidemiology 2003;22:217-228.
9. Rose MR. Gulf War service is an uncertain trigger for ALS. Neurology 2003; 61:730-731.

Reply to Brenner and Armon 27 October 2003

Ronnie Horner, PhD,
NIH/NINDS
NINDS, NSC Building�Rm 2153, 6001 Executive Blvd, Bethesda, MD 20852,
Ronnie D. Horner, PhD and John R. Feussner, MD, MPH for the Investigators of VA Cooperative Studies Program Project #500
Send Correspondence to journal:
Re: Reply to Brenner and Armon

hornerr{at}ninds.nih.gov Ronnie Horner, PhD, et al.

We appreciate the interest in our study and understand the criticism of our conclusion that 1991 Gulf War veterans experience a two-fold elevated risk of amyotrophic lateral sclerosis.
We agree with Dr. Brenner that investigators need to consider the full array of possible etiological factors in follow-up studies and so indicated in our article. [1]
Dr. Armon succinctly presents the major criticism of our study--that the findings are explained by under-ascertainment of cases among the non-deployed population. The choice of the comparison population is at the core of the criticism. As an occupational epidemiologic study, the most appropriate primary comparison group is the non-deployed military population because it is the most similar to the deployed population in terms of selection for service, that is, health and physical fitness, and other factors. Given the selective nature of the military population, rates lower than observed in a general population are expected. Moreover, the level of under-ascertainment in the non-deployed would have to be substantial to cancel out the elevation in risk. That scenario lacks a compelling basis. Our capture-recapture analysis that used methods appropriate when source dependency exists, indicated an elevated rate after adjustment for the likely level of under- ascertainment. [2,3] If we adjust for under-ascertainment sufficient to eliminate the elevated risk, the age-adjusted rate among the non-deployed exceeds that of the general population.
Dr. Armon also speculates that our findings could be confounded by smoking or other factors. To confound the association, smoking and the other factors would have to be substantially more prevalent among the deployed than non-deployed military population, an unproven and unlikely assumption. As for confounding by the incubation period of ALS, when we excluded cases with onset prior to July, 1991, an elevated risk persisted (Table 4). The lack of a biologically plausible hypothesis for the study is not relevant. Our purpose was to determine if an association existed; etiology is sought only after an association is identified. [4] Moreover, it is well recognized that reliance on biological plausibility and other such criteria to infer causative relationships is highly vulnerable to erroneous conclusions. [5]
We are disheartened with the premature closure of thought regarding the possible value of our findings to understanding the etiology of ALS. We admit that this cluster may not identify the etiology of ALS and it may suggest a direction for further studies. We advocate open- minded, creative thinking about possible factors.
References
1. Horner RD, Kamins KG K, Feussner JR et al. Occurrence of amyotrophic lateral sclerosis among Gulf War veterans. Neurology 2003; 61: 742-9.
2. Hook EB, Regal RR. Capture-recapture methods in epidemiology: methods and limitations. Epidemiol. Rev 1995; 17:243-264.
3. Chao A., Tsay PK, Lin S, Shau W, Chao D. The application of capture-recapture models to epidemiological data. Statist. Med. 2001; 20:3123-3157.
4. Gregg MB. Field Epidemiology. New York: Oxford University Press, 1996.
5. Rothman KJ, Greenland S. Modern Epidemiology, second edition. Philadelphia: Lippencott-Raven, 1998.