Tom Kindlon
Senior Member (Voting Rights)
I don't think some of these findings were ever published.
I'm clearing out some of the mass of paper I have and thought I should highlight it somewhere
https://listserv.nodak.edu/cgi-bin/wa.exe?A2=ind0910C&L=CO-CURE&P=R2735
CFSCC meeting, on April 22 (1999).
DR. REEVES: The objectives of the CDC's chronic fatigue
syndrome program are to: determine the pathogenesis of CFS, testing
the hypothesis that the syndrome represents the outcome of several
unrelated initial insults; to estimate the magnitude of the problem,
that is to say, prevalence and incidence in the Unite States; to
define the natural history of CFS; to identify risk factors and
diagnostic markers; to provide current, appropriate technical
information on CFS to government agencies, public health officials,
health care providers, patients, and the public; and ultimately, to
develop control strategies.
Within the CFS Program we attempt to address these objectives
within the context of patient concerns. And current patient concerns
of highest priority, with particular relevance to CDC's CFS Program,
include: patient care -- that's diagnosis and detection, treatment,
rehabilitation, appropriate "third-party" coverage; to conduct
research to identify the cause of illness and diagnostic markers;
estimates of the prevalence and incidence of CFS; a revision of the
case definition and the name; and definition of CFS characteristics
-- prevalence and clinical features -- in underserved populations,
particular interest on children and racial-ethnic minorities.
What I want to do in this talk is to summarize where we've been
since I last presented. We're doing a large study in Cedric County,
which is Wichita. The primary objective of this study is to estimate
the prevalence of CFS and other fatiguing illnesses in the Wichita
population. We selected Wichita in part because it is generally
representative of the United States.
There are four secondary objectives. First, we follow study
participants annually to better understand the clinical course of CFS
and to estimate the incidence of CFS and other fatiguing illnesses.
We use information collected from study participants to explore
empirically derived case definitions. Clinical specimens from study
subjects are subjected to various laboratory assays. And finally,
this study will allow comparison of population-based prevalence
estimates with those previously derived from the physician-based
surveillance.
I want to address the baseline survey & clinical evaluation.
The baseline survey was a 3-stage investigation composed of screening
and detailed telephone interviews followed by clinical evaluation.
From March through August of '97, random digit dialing was used to
screen households and obtain information on fatiguing illness.
Approximately 90,000 people were enumerated, which represents about a
quarter of Wichita's population, so this was a very large
population-based study. Approximately 6,500 people who reported
fatigue lasting a month or longer and a similar number of
non-fatigued controls were asked to complete a detailed telephone
interview.
Finally, the clinical evaluations were performed on 300
subjects who reported symptoms which were criteria for CFS and when
we detect someone who meets the case definition on interview, we call
that CFS-like illness, and 64 non-fatigued subjects.
At the October, 1998 meeting, I reported preliminary CFS
prevalence estimates. At that time, we had sufficient data to
classify 39 of 300 clinically evaluated fatigued subjects as CFS. We
have now completed follow-up clinical studies on subjects that were
left as pending at first clinical evaluation, we really did not know
whether to rule them out or not, and we've identified a total of 49
CFS patients at base-line.
We used statistical methods to account for our sampling design,
and our current estimates are that 238 per 100,000 adults 18-69 years
of age in Wichita have CFS. This translates into about 700 cases in
Cedric county. Women accounted for the majority of cases, with a
prevalence of 356 per 100,000; and most of these cases occurred in
white women, resulting in a prevalence of 394/100,000. This is close
to a half of a percent. These are very, very high numbers,
significantly higher than we had thought in the past.
With respect to adolescent prevalence, the telephone survey
also obtained information on fatiguing illness in adolescents.
Because of limitations imposed by Human Subjects Committee
considerations, we could only conduct detailed interviews and
clinical evaluations on adolescents 12-17. Seven adolescents with
CFS-like illness were clinically evaluated and none had CFS.
However, -- so it is very hard to calculate a prevalence.
However, telephone interview data and US census estimates
allowed us to estimate the prevalence of CFS-like illness in children
or in adolescents -- and that was 338 per 100,000. This is about a
fifth the rate of CFS-like illness in adults, which was 1,623 per
100,000. If we assume that the proportion of CFS to CFS-like found
in the adult population, which was 15%, could be applied to
adolescents, we would estimate -- or we do estimate now that 50 per
100,000 adolescents 12-17 years of age in Wichita have CFS.
The first year of follow-up study occurred between February and
October of last year. The objective was to collect updated
information on the health status of the 13,000 first-phase
respondents.
Used a telephone interview which was very similar, somewhat modified
from the baseline one. All subjects originally invited to the clinic
at baseline, as well as those newly identified with CFS-like illness
were asked to participate in the second-phase clinical evaluation.
And we are currently calculating incidence rates of CFS and other
fatiguing illnesses.
With respect to current work. Final analysis of Wichita
prevalence data will be completed in this year and a manuscript will
be submitted for publication. Analysis of the data with respect to
empirically deriving case definitions will be completed this year and
a manuscript will be submitted for publication. We want to submit it
to a high-visibility, weekly journal.
Analysis of the data with respect to empirically deriving case
definitions -- case definitions from the data is underway. We will
complete that this year and a manuscript will be submitted for
publication.
Twenty-four month follow-up began in February and is ongoing.
It will be completed this year and we will make decision concerning
continuation for three years.
We have a variety of new analyses that we are actively engaged
in. We obtained a wealth of data from the Wichita study and we are
undertaking a variety of other analyses. The subjects of major
importance to us right now include: an analysis of the burden of CFS
on the population. We talked a lot about this yesterday. We talked
about continuing medical education. Only 9 -- and that's 18 percent
-- of the 49 subjects with CFS that we identified, reported they had
been treated or diagnosed with this illness. It will be extremely
important to describe the utilization of health services by
individuals with fatiguing illnesses.
The relationships between CFS-like and CFS. In all, 1,600 per
100,000 adults were identified with CFS-like illness during a
telephone survey, yet 15 percent, 238 per 100,000 had CFS confirmed
upon standard clinical evaluation. To some extent, that reflects
extremely sensitive screening instruments with low specificity, which
is good. But we are going to need to look at this in detail for some
of the future studies.
Differences in responses obtained in telephone screening and
clinical interview. We have found several instances in which
subjects provided different responses to the detailed telephone
interview and the in-person interview conducted at the clinic. We
are exploring that in detail.
We are very interested in the comparison of passive versus
active surveillance for CFS. Again, this was brought up yesterday,
perhaps, in wanting physicians to report to us. Now we had done that
previously from
'89 to '93. Wichita-based prevalence rates are more than 20 times
higher than those estimated through physician surveillance, and we're
conducting an analysis of that. Interestingly, 15 -- or the 18
percent of patients who sought physicians, if we calculate prevalence
based on that, we calculate exactly the same prevalence we calculated
from physician-based surveillance.
We have hired two epidemiologists to work on these various
analyses and we are advertising a post-doctoral fellowship that we
hope the fill later this year. Those who are academics on this panel
we would invite any suggestions or perhaps people wishing to take
sabbatical at CDC.
We have considered two applicants for the neuroendocrinologist
position. Both had excellent laboratory backgrounds but lacked
adequate clinical and epidemiologic expertise. And we are continuing
to search for a qualified neuroendocrinologist or neuroimmunologist
with epidemiologic expertise.
I'd like to talk about the new study that we are planning. We
are planning a national survey for CFS in the US which we hope to
begin this year. The objective will be to conduct a telephone survey
to estimate sex, age, race, and socioeconomic-specific
prevalence of medically unexplained fatigue, prolonged fatigue of one
to five months; chronic fatigue, one to six months without syndromic
features; and CFS-like illness. Special emphasis in this survey will
be given to identifying fatiguing illness in adolescents and
racial/ethnic minorities.
Specific aims are to determine if findings from the Wichita
study can be generalized to the US population; to estimate the
geographic occurrence of prolonged fatigue, chronic fatigue, and
chronic fatigue syndrome-like illness. In particular, are there
metropolitan, urban, rural differences? Are there
north/south/east/west differences or any indication of cluster. Are
there differences associated with migration? That sort of
information.
Collect information that can be used to verify, complement, and
extend empirically derived case definitions. Collect information
that can be used to estimate the economic burden of CFS-like illness,
which would include utilization of health services with respect to
socioeconomic status or occupation as well as changes in
socioeconomic status/occupation due to illness.
Derive information that can be used in future case control
studies, which would include identifying subjects for future recall,
identifying high-risk areas, and deriving hypotheses for analytic
studies.
And finally, to derive information for use in designing a
national or regional CFS registry.
I'd like to quickly cover what we have been doing in molecular
epidemiologic analysis. We discussed this last time. Classic case
control studies have not consistently identified laboratory markers
or risk factors for CFS. We believe that an open-ended analysis of
differences in gene expression between CFS patients and controls will
give the best opportunity for providing insight into disease -- this
sort of disease with an unknown pathogenesis.
Our initial approach uses high-density filter arrays to
identify differences in gene expression in peripheral blood
lymphocytes, white cells. This is an extremely new technology which
has not yet been applied to epidemiologic studies, and has been
applied very rarely so far to studies of unknown things. Thus we
have had to carefully standardize our assays. I discussed this last
time. We have now optimized sample collection, storage methods, and
labeling methods for chemiluminescent analysis of gene expression.
We have used the assay in relatively simple tissue culture
systems infected with human papillomavirus or HPV, to standardize
both the technique and the analysis methods. HPV has only 7 genes
and their functions are well described. These genes can be up or
down regulated, turned off, on or down, depending on whether the
virus exists free in a free state in the cell, which is called an
episomal state, or whether the viral DNA has fused itself with the
human DNA, which is called an integrated state.
We have measured cellular messenger RNA expression in this
system and are currently utilizing different mathematical approaches
to identify and characterize the differences, and I'm going to show
some of those quickly.
We have also begun to test archived samples from the 1993
Atlanta Case Control Study -- we've got six of them done so far.
We're testing them against four different array formats. So we're
testing gene expressing in a Stress Array, in a Neurobiology Array,
in an Immunology Array, and in a Cytokine Array. Each one of these
tests 588 different genes. We're now looking at new formats that
will test 5000 to 10,000 genes.
What I'm going to do is just -- with some apologies which I
hate to do, for the technical nature of this -- Doctors Komaroff and
Klimas raised some doubts which we shared. This is how this study is
done. A piece of filter paper about half the size of an 8x10 sheet
of paper is used. It has six quadrants. This is just a general
array. One quadrant has onco-genes, two are suppressor genes, stress
response genes, apoptosis which is programmed cell killing, et
cetera. This is just a general filter.
The way in which this is done is that RNA is extracted from the
cells of interest, copies are made in the DNA format. These are
hybridized or put on the filters and allowed to combine and the
filters are then looked at using photographic film.
And this is what a filter looks like. I showed you this one, I
think, last time. This is from our HPV studies. This is normal
cells. These are cells infected with HPV in an episomal form. You
can see that for example, right here, this is very dark. It's not
very dark right there. It's more -- these are exactly the same
except for the two things -- you can see the differences in
expression. This doesn't help anybody at all. It's kind of neat and
I can give a nice talk about it. We can, in fact, quantify it.
And this is an example of the output quantifying this. So for
example, episomal reinfected cells are expressing gene -- a gene in
quadrant A-1b, A-1a, normal cells are not. Here there's the same
amount of expression -- this is the kind of output we get for 588
genes on each filter.
Now, how can we tell the differences in these? There are two
ways, and again, I apologize for this. It looked really, really neat
when I did it on my computer. Doesn't look quite so neat here. This
is the ratio of signal in an episomally-infected cell compared to a
normal cell, and you can see how the genes fall out there. Now what
you do is you look for large differences in expression. And so we
can take those genes that are expressed by episomal but not by forced
enkarytinocytes (ph) and look at each one of those.
We can look at those and you can call these CFS patients, CFS
patients derived by a numeric case definition, sudden onset, slow
onset, et cetera. There would be genes expressed only by those
patients and not by controls; there would be genes that are expressed
by both that are expressed much more by the cases than the controls;
or by the controls than the cases. And one needs to look at those
genes and see what they're doing. This may lead us to hypotheses of
pathogenesis.
As far as a diagnostic marker which I think is much more
exciting or immediate use, we can also compare the overall profiles
of these genes and determine whether we can see differences
mathematically in populations defined by the profiles. And you can
see just looking at these profiles of cell cycle/cell response et
cetera gene, that at least within the known system, there are quite
easily demonstrable differences. This type of analysis could be used
to look at and differentiate between those sorts of systems, and
potentially to determine -- and again, obviously, this is a
population of cells -- if this is reproducible between patients to
determine between patients and non patients.
More interesting yet, and this is coming from a rapidly
evolving technology, mathematical models exist using parsimony
analysis to derive family trees, and this is looking at -- we had no
reason to believe that this would work, except that it did, and it
fit our hypothesis. This is mathematical model of the output of two
arrays which can very clearly distinguish non-infected cells in a
branch of the family tree, cells which have unregulated HPV
expression because of integrated DNA, and intermediate populations
which is due to -- using technical jargon -- multiply integrated DNA
and episomal DNA.
And what we hope to do is to use this sort of analysis, which
is quite different actually. The more genes and the more people you
have, the greater likelihood you have to do it, to derive algorithms
for clearly distinguishing patients with chronic fatigue syndrome or
other fatiguing illness, based on overall gene expression.
DR. KLIMAS: Is there a factor analysis?
DR. REEVES: I beg your pardon?
(continues)
I'm clearing out some of the mass of paper I have and thought I should highlight it somewhere
https://listserv.nodak.edu/cgi-bin/wa.exe?A2=ind0910C&L=CO-CURE&P=R2735
CFSCC meeting, on April 22 (1999).
DR. REEVES: The objectives of the CDC's chronic fatigue
syndrome program are to: determine the pathogenesis of CFS, testing
the hypothesis that the syndrome represents the outcome of several
unrelated initial insults; to estimate the magnitude of the problem,
that is to say, prevalence and incidence in the Unite States; to
define the natural history of CFS; to identify risk factors and
diagnostic markers; to provide current, appropriate technical
information on CFS to government agencies, public health officials,
health care providers, patients, and the public; and ultimately, to
develop control strategies.
Within the CFS Program we attempt to address these objectives
within the context of patient concerns. And current patient concerns
of highest priority, with particular relevance to CDC's CFS Program,
include: patient care -- that's diagnosis and detection, treatment,
rehabilitation, appropriate "third-party" coverage; to conduct
research to identify the cause of illness and diagnostic markers;
estimates of the prevalence and incidence of CFS; a revision of the
case definition and the name; and definition of CFS characteristics
-- prevalence and clinical features -- in underserved populations,
particular interest on children and racial-ethnic minorities.
What I want to do in this talk is to summarize where we've been
since I last presented. We're doing a large study in Cedric County,
which is Wichita. The primary objective of this study is to estimate
the prevalence of CFS and other fatiguing illnesses in the Wichita
population. We selected Wichita in part because it is generally
representative of the United States.
There are four secondary objectives. First, we follow study
participants annually to better understand the clinical course of CFS
and to estimate the incidence of CFS and other fatiguing illnesses.
We use information collected from study participants to explore
empirically derived case definitions. Clinical specimens from study
subjects are subjected to various laboratory assays. And finally,
this study will allow comparison of population-based prevalence
estimates with those previously derived from the physician-based
surveillance.
I want to address the baseline survey & clinical evaluation.
The baseline survey was a 3-stage investigation composed of screening
and detailed telephone interviews followed by clinical evaluation.
From March through August of '97, random digit dialing was used to
screen households and obtain information on fatiguing illness.
Approximately 90,000 people were enumerated, which represents about a
quarter of Wichita's population, so this was a very large
population-based study. Approximately 6,500 people who reported
fatigue lasting a month or longer and a similar number of
non-fatigued controls were asked to complete a detailed telephone
interview.
Finally, the clinical evaluations were performed on 300
subjects who reported symptoms which were criteria for CFS and when
we detect someone who meets the case definition on interview, we call
that CFS-like illness, and 64 non-fatigued subjects.
At the October, 1998 meeting, I reported preliminary CFS
prevalence estimates. At that time, we had sufficient data to
classify 39 of 300 clinically evaluated fatigued subjects as CFS. We
have now completed follow-up clinical studies on subjects that were
left as pending at first clinical evaluation, we really did not know
whether to rule them out or not, and we've identified a total of 49
CFS patients at base-line.
We used statistical methods to account for our sampling design,
and our current estimates are that 238 per 100,000 adults 18-69 years
of age in Wichita have CFS. This translates into about 700 cases in
Cedric county. Women accounted for the majority of cases, with a
prevalence of 356 per 100,000; and most of these cases occurred in
white women, resulting in a prevalence of 394/100,000. This is close
to a half of a percent. These are very, very high numbers,
significantly higher than we had thought in the past.
With respect to adolescent prevalence, the telephone survey
also obtained information on fatiguing illness in adolescents.
Because of limitations imposed by Human Subjects Committee
considerations, we could only conduct detailed interviews and
clinical evaluations on adolescents 12-17. Seven adolescents with
CFS-like illness were clinically evaluated and none had CFS.
However, -- so it is very hard to calculate a prevalence.
However, telephone interview data and US census estimates
allowed us to estimate the prevalence of CFS-like illness in children
or in adolescents -- and that was 338 per 100,000. This is about a
fifth the rate of CFS-like illness in adults, which was 1,623 per
100,000. If we assume that the proportion of CFS to CFS-like found
in the adult population, which was 15%, could be applied to
adolescents, we would estimate -- or we do estimate now that 50 per
100,000 adolescents 12-17 years of age in Wichita have CFS.
The first year of follow-up study occurred between February and
October of last year. The objective was to collect updated
information on the health status of the 13,000 first-phase
respondents.
Used a telephone interview which was very similar, somewhat modified
from the baseline one. All subjects originally invited to the clinic
at baseline, as well as those newly identified with CFS-like illness
were asked to participate in the second-phase clinical evaluation.
And we are currently calculating incidence rates of CFS and other
fatiguing illnesses.
With respect to current work. Final analysis of Wichita
prevalence data will be completed in this year and a manuscript will
be submitted for publication. Analysis of the data with respect to
empirically deriving case definitions will be completed this year and
a manuscript will be submitted for publication. We want to submit it
to a high-visibility, weekly journal.
Analysis of the data with respect to empirically deriving case
definitions -- case definitions from the data is underway. We will
complete that this year and a manuscript will be submitted for
publication.
Twenty-four month follow-up began in February and is ongoing.
It will be completed this year and we will make decision concerning
continuation for three years.
We have a variety of new analyses that we are actively engaged
in. We obtained a wealth of data from the Wichita study and we are
undertaking a variety of other analyses. The subjects of major
importance to us right now include: an analysis of the burden of CFS
on the population. We talked a lot about this yesterday. We talked
about continuing medical education. Only 9 -- and that's 18 percent
-- of the 49 subjects with CFS that we identified, reported they had
been treated or diagnosed with this illness. It will be extremely
important to describe the utilization of health services by
individuals with fatiguing illnesses.
The relationships between CFS-like and CFS. In all, 1,600 per
100,000 adults were identified with CFS-like illness during a
telephone survey, yet 15 percent, 238 per 100,000 had CFS confirmed
upon standard clinical evaluation. To some extent, that reflects
extremely sensitive screening instruments with low specificity, which
is good. But we are going to need to look at this in detail for some
of the future studies.
Differences in responses obtained in telephone screening and
clinical interview. We have found several instances in which
subjects provided different responses to the detailed telephone
interview and the in-person interview conducted at the clinic. We
are exploring that in detail.
We are very interested in the comparison of passive versus
active surveillance for CFS. Again, this was brought up yesterday,
perhaps, in wanting physicians to report to us. Now we had done that
previously from
'89 to '93. Wichita-based prevalence rates are more than 20 times
higher than those estimated through physician surveillance, and we're
conducting an analysis of that. Interestingly, 15 -- or the 18
percent of patients who sought physicians, if we calculate prevalence
based on that, we calculate exactly the same prevalence we calculated
from physician-based surveillance.
We have hired two epidemiologists to work on these various
analyses and we are advertising a post-doctoral fellowship that we
hope the fill later this year. Those who are academics on this panel
we would invite any suggestions or perhaps people wishing to take
sabbatical at CDC.
We have considered two applicants for the neuroendocrinologist
position. Both had excellent laboratory backgrounds but lacked
adequate clinical and epidemiologic expertise. And we are continuing
to search for a qualified neuroendocrinologist or neuroimmunologist
with epidemiologic expertise.
I'd like to talk about the new study that we are planning. We
are planning a national survey for CFS in the US which we hope to
begin this year. The objective will be to conduct a telephone survey
to estimate sex, age, race, and socioeconomic-specific
prevalence of medically unexplained fatigue, prolonged fatigue of one
to five months; chronic fatigue, one to six months without syndromic
features; and CFS-like illness. Special emphasis in this survey will
be given to identifying fatiguing illness in adolescents and
racial/ethnic minorities.
Specific aims are to determine if findings from the Wichita
study can be generalized to the US population; to estimate the
geographic occurrence of prolonged fatigue, chronic fatigue, and
chronic fatigue syndrome-like illness. In particular, are there
metropolitan, urban, rural differences? Are there
north/south/east/west differences or any indication of cluster. Are
there differences associated with migration? That sort of
information.
Collect information that can be used to verify, complement, and
extend empirically derived case definitions. Collect information
that can be used to estimate the economic burden of CFS-like illness,
which would include utilization of health services with respect to
socioeconomic status or occupation as well as changes in
socioeconomic status/occupation due to illness.
Derive information that can be used in future case control
studies, which would include identifying subjects for future recall,
identifying high-risk areas, and deriving hypotheses for analytic
studies.
And finally, to derive information for use in designing a
national or regional CFS registry.
I'd like to quickly cover what we have been doing in molecular
epidemiologic analysis. We discussed this last time. Classic case
control studies have not consistently identified laboratory markers
or risk factors for CFS. We believe that an open-ended analysis of
differences in gene expression between CFS patients and controls will
give the best opportunity for providing insight into disease -- this
sort of disease with an unknown pathogenesis.
Our initial approach uses high-density filter arrays to
identify differences in gene expression in peripheral blood
lymphocytes, white cells. This is an extremely new technology which
has not yet been applied to epidemiologic studies, and has been
applied very rarely so far to studies of unknown things. Thus we
have had to carefully standardize our assays. I discussed this last
time. We have now optimized sample collection, storage methods, and
labeling methods for chemiluminescent analysis of gene expression.
We have used the assay in relatively simple tissue culture
systems infected with human papillomavirus or HPV, to standardize
both the technique and the analysis methods. HPV has only 7 genes
and their functions are well described. These genes can be up or
down regulated, turned off, on or down, depending on whether the
virus exists free in a free state in the cell, which is called an
episomal state, or whether the viral DNA has fused itself with the
human DNA, which is called an integrated state.
We have measured cellular messenger RNA expression in this
system and are currently utilizing different mathematical approaches
to identify and characterize the differences, and I'm going to show
some of those quickly.
We have also begun to test archived samples from the 1993
Atlanta Case Control Study -- we've got six of them done so far.
We're testing them against four different array formats. So we're
testing gene expressing in a Stress Array, in a Neurobiology Array,
in an Immunology Array, and in a Cytokine Array. Each one of these
tests 588 different genes. We're now looking at new formats that
will test 5000 to 10,000 genes.
What I'm going to do is just -- with some apologies which I
hate to do, for the technical nature of this -- Doctors Komaroff and
Klimas raised some doubts which we shared. This is how this study is
done. A piece of filter paper about half the size of an 8x10 sheet
of paper is used. It has six quadrants. This is just a general
array. One quadrant has onco-genes, two are suppressor genes, stress
response genes, apoptosis which is programmed cell killing, et
cetera. This is just a general filter.
The way in which this is done is that RNA is extracted from the
cells of interest, copies are made in the DNA format. These are
hybridized or put on the filters and allowed to combine and the
filters are then looked at using photographic film.
And this is what a filter looks like. I showed you this one, I
think, last time. This is from our HPV studies. This is normal
cells. These are cells infected with HPV in an episomal form. You
can see that for example, right here, this is very dark. It's not
very dark right there. It's more -- these are exactly the same
except for the two things -- you can see the differences in
expression. This doesn't help anybody at all. It's kind of neat and
I can give a nice talk about it. We can, in fact, quantify it.
And this is an example of the output quantifying this. So for
example, episomal reinfected cells are expressing gene -- a gene in
quadrant A-1b, A-1a, normal cells are not. Here there's the same
amount of expression -- this is the kind of output we get for 588
genes on each filter.
Now, how can we tell the differences in these? There are two
ways, and again, I apologize for this. It looked really, really neat
when I did it on my computer. Doesn't look quite so neat here. This
is the ratio of signal in an episomally-infected cell compared to a
normal cell, and you can see how the genes fall out there. Now what
you do is you look for large differences in expression. And so we
can take those genes that are expressed by episomal but not by forced
enkarytinocytes (ph) and look at each one of those.
We can look at those and you can call these CFS patients, CFS
patients derived by a numeric case definition, sudden onset, slow
onset, et cetera. There would be genes expressed only by those
patients and not by controls; there would be genes that are expressed
by both that are expressed much more by the cases than the controls;
or by the controls than the cases. And one needs to look at those
genes and see what they're doing. This may lead us to hypotheses of
pathogenesis.
As far as a diagnostic marker which I think is much more
exciting or immediate use, we can also compare the overall profiles
of these genes and determine whether we can see differences
mathematically in populations defined by the profiles. And you can
see just looking at these profiles of cell cycle/cell response et
cetera gene, that at least within the known system, there are quite
easily demonstrable differences. This type of analysis could be used
to look at and differentiate between those sorts of systems, and
potentially to determine -- and again, obviously, this is a
population of cells -- if this is reproducible between patients to
determine between patients and non patients.
More interesting yet, and this is coming from a rapidly
evolving technology, mathematical models exist using parsimony
analysis to derive family trees, and this is looking at -- we had no
reason to believe that this would work, except that it did, and it
fit our hypothesis. This is mathematical model of the output of two
arrays which can very clearly distinguish non-infected cells in a
branch of the family tree, cells which have unregulated HPV
expression because of integrated DNA, and intermediate populations
which is due to -- using technical jargon -- multiply integrated DNA
and episomal DNA.
And what we hope to do is to use this sort of analysis, which
is quite different actually. The more genes and the more people you
have, the greater likelihood you have to do it, to derive algorithms
for clearly distinguishing patients with chronic fatigue syndrome or
other fatiguing illness, based on overall gene expression.
DR. KLIMAS: Is there a factor analysis?
DR. REEVES: I beg your pardon?
(continues)