(CANCER RESEARCH 49. 4038-4041. July 15. 1989)
Infectious and Noninfectious Exposures in the Etiology of Light Chain Myeloma: A
Case-Control Study1
Ann R. Williams,2 Noel S. Weiss, Thomas D. Koepsell, Joseph L. Lyon, and G. Marie Swanson
Department of Epidemiology, University of Washington, Seattle 98195 [A. R. W., N. S. W., T. D. K.J; Fred Hutchinson Cancer Research Center, Seattle, Washington
98104 [A. R. H'., N. S. W., T. D. K.¡;Division of Epidemiology, Department of Family and Community Medicine, University of Utah Medical Center, Salt Lake City,
Utah 84132 ¡J.L. LJ; and Michigan Cancer Foundation, Division of Epidemiology, Detroit, Michigan 48200 ¡C.M. S.]
ABSTRACT
Sixty-nine subjects with light chain myeloma were interviewed in a
multicenter case-control study, and their responses were compared to
those of 1683 controls selected from the general population of the same
geographic areas. The interview was directed at the subject's history of
exposure to a variety of chemical and infectious agents. Persons with a
history of a medical implant had 2.2 times the risk of other persons (95%
confidence interval = 0.9-5.8), a relative risk that increased with increas
ing time that the implant had been present. Alkali exposure that was
deemed by the subject to be unusually heavy was associated with a
relative risk of light chain myeloma of 7.8 (95% confidence interval =
1.7-35.3), while similarly defined exposure to carbon monoxide increased
the risk by 6.1 times (95% confidence interval = 2.0-18.2). These findings
differ from those obtained in our study of the more common forms of
multiple myeloma and, while the differences are plausibly due only to
chance given the large number of exposures investigated, they could be
an indication that light chain myeloma is an etiologically distinct entity.
INTRODUCTION
Light chain myeloma, representing about 15% of all multiple
myelomas, is a form of the disease that is characterized by
monoclonal light chain production but no evidence of produc
tion of intact monoclonal immunoglobulin. Patients with light
chain myeloma are clinically distinct from those with idiopathic
Bence Jones proteinuria, as they have other evidence of active
multiple myeloma (1). The difference in pathogenesis between
light chain myeloma and the more common forms of the disease
is not known. Some have postulated that in the case of light
chain myeloma, the malignant clone is derived from a less
differentiated cell, incapable of manufacturing intact immuno
globulin. This analysis was performed to examine the differ
ences in exposure history between light chain myeloma patients
and a suitable control group and to compare these results with
those obtained in a similar analysis of all forms of multiple
myeloma combined.
SUBJECTS
AND METHODS
counties comprising metropolitan Atlanta. Subjects who were under 80
years old and diagnosed between July 1, 1977 and June 30, 1981 were
eligible (n = 773). Two thousand twenty-four controls were drawn from
the general population of the same areas. Interview data were obtained,
either from the subjects themselves or from their next of kin, on 89%
of cases (698 subjects) and 83% of identified controls (1683) subjects).
An average of 1 year elapsed between the time a case was diagnosed
and when the interview took place.
Clinical Data. Hospital and outpatient records were abstracted to
obtain the results of serum protein electrophoresis and immunoelectrophoresis, bone survey, urine protein electrophoresis, and/or BenceJones protein determination and percentage of plasmacytosis on bone
marrow aspiration/biopsy. Records were available for 85% of cases.
Case Definition. Subjects were classified as having light chain mye
loma if there was a positive test for Bence Jones proteinuria and/or a
positive urine protein electrophoresis, but (a) the immunoelectrophoresis was negative, and (/>)the serum protein electrophoresis revealed
either no monoclonal spike or a spike that was less than 0.5 g/dl.
Thirteen cases were included in whom there was insufficient informa
tion available to fulfill the above-mentioned criteria, but the medical
record abstractor found the diagnosis of light chain myeloma explicitly
stated in the patients' records. In every instance in which records were
available, confirmatory evidence supporting the diagnosis of multiple
myeloma was found in either bone marrow aspirates or on skeletal
survey or both. In most cases, IgD and IgE myelomas were not specif
ically excluded. Given the rarity of these subtypes, the probability that
such patients were erroneously included is small.
Interview Data. Trained interviews administered a standardized ques
tionnaire focusing on factors hypothesized to be etiologically related to
the development of multiple myeloma (e.g., chronic antigenic stimula
tion, allergies, drug histories, radiation, and chemical exposure), in
addition to basic demographic data. With respect to chemical expo
sures, the pertinent question read as follows: "In your daily life, at
home, at work, or elsewhere, was there ever a time when you were
highly exposed to products or fumes such as gasoline, turpentine,
alcohol (other than liquor), cleaning solvents, oil, or other (specify)."
Due to illness and death, only 67% of case interviews were obtained
directly from the subject, while 33% of case interviews were obtained
from a next-of-kin proxy. Ninety-nine% of control interviews were
obtained from the subject.
Formation of Chemical Exposure Categories. Without knowledge of
case-control status, all responses to "other (specify)" in the chemical
exposure section were hand tabulated and grouped into 20 exposure
categories with the aid of a toxicologist. (See "Appendix" for a list of
Results reported here are based upon a large multicenter collabora
tive case-control study of multiple myeloma. Details of this study have
been previously published (2). Briefly, using records of populationbased cancer registries, persons with newly diagnosed multiple myeloma
were identified from four geographic areas: (a) King and Pierce Coun
ties in Washington; (b) Davis, Salt Lake, and Weber Counties in Utah;
(c) the three counties comprising metropolitan Detroit; and (d) the five
major substances included in each category.) Affirmative responses to
the substances gasoline, cleaning solvents, benzene, and oil were in
cluded in the aliphatic hydrocarbon, chlorinated hydrocarbon, aromatic
hydrocarbon, and oil exposure categories, respectively. Certain re
sponses indicating exposure to more than one chemical were placed in
several categories.
Analysis. Possible differences between cases and controls were con
trolled for age, sex, race, study area, and educational attainment by
means of the Mantel-Haenszel procedure (3). The 95% confidence
Received 7/21/88; revised 2/6/89; accepted 4/11/89.
interval around the relative risk estimated by that procedure was cal
The costs of publication of this article were defrayed in part by the payment
of page charges. This article must therefore be hereby marked advertisement in
culated by the method of Miettinen (4). For odds ratios of zero, the
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
95% confidence interval was calculated by the exact method (5). Sepa
' Supported in part by Grant 1 NOI CA 23350 from the National Cancer
rate analyses were performed, first using all cases on whom exposure
Institute.
1To whom requests for reprints should be addressed, at Fred Hutchinson
information was available, and then using only those cases who them
Cancer Research Center, 1124 Columbia Street (W 108), Seattle. WA 98104.
selves were interviewed.
4038
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1989 American Association for Cancer Research.
LIGHT CHAIN MYELOMA: CASE-CONTROLSTUDY
Table 1 Risk of light chain myeloma in relation to noninfectious exposure
% reporting exposure
Adjusted relative risk
CasesExposureChildhood
respondentsRR°2.10.70.70.20.61.12.31.54.02.50.095%
69)4.321.44.32.931.428.611.44.37.12.9Self(n
=
46)6.329.84.32.129.831.910.64.26.42.1Controls(n=
=
1683)1.729.07.45.744.924.77.02.74.41.72.1All
CI0.7-6.70.4-1.40.2-2.50.0-1.70.3-1.00.6-1.91.1-5.00.5-4.51.5-10.30.5-12.20.0
Cl0.8-8.40.6-2.40.2-3.10.1-2.70.3-1.00.6-2.
eczemaAllergyAllergy
shotsAsthma*Arthritis'Blood
transfusionMedical
implantDuration
implant<5yr>5yrShrapnelSerum
of
sickness''All(n
°RR, risk relative to that of persons with no history of that exposure, adjusted for age, sex, race, residence, and educational attainment.
* Relative risk in self-respondents, adjusted for age, sex, and race.
' Self-reported history of arthritis (type unspecified) and without medical confirmation.
d Exact confidence interval, adjusted for age and race.
Table 2 Distribution of implant location and material among those reporting a
history of medical implant
RESULTS
exposureImplant
Cases and controls differed with respect to their histories of
some noninfectious antigenic exposures (Table 1). Three cases
reported a history of childhood eczema, whereas only one case
would have been expected, based on the experience among
controls. However, this difference was well within the bounds
of chance. Cases had positive histories of asthma and "arthritis"
% reporting
(n 8)631324502525Controls
=
<n=141)501634642620
categoryLocationOrthopedicIntravascularOtherMaterialMetalPlasticOtherCases
less often than controls, but these associations also were easily
consistent with there being no true differences. A greater pro
portion of cases than controls reported a history of a foreign
body implanted for medical reasons, with an adjusted odds ratio
(self-respondents) of 2.2 (95% CI' = 0.9-5.8). The risk associ
ated with exposure to medical implants increased with increas
ing duration of exposure, the relative risk increasing from 1.5
(95% CI = 0.3-6.4) among these exposed 5 years or less, to 3.5
(95% CI = 1.1-11.4) among those exposed for greater than 5
years. Among those with implants, neither the distribution of
type nor location differed between cases and controls (Table 2).
With few exceptions, cases and controls reported a history of
various types of infection with comparable frequency (Table 3).
A history of chickenpox was associated with a relative risk of
0.5 (95% CI = 0.3-1.0). However, the odds ratio varied consid
erably with age. Only among subjects 50 to 59 years old did the
odds ratio (0.1) suggest a truly protective effect. Among the
other age groups, the odds ratio varied between 0.8 and 1.1,
suggesting no protective effect of chickenpox infection. A his
tory of fever blisters and scarlet fever were also reported less
often by cases, but these differences could easily have occurred
by chance.
There was no evidence of a relationship between light chain
myeloma and diagnostic radiation (Table 4).
Among the 20 chemical groups reported in Table 5, exposure
to alkali and other caustics, asbestos, carbon monoxide, metals,
and pesticides were all associated with odds ratios over 2.0. For
most of these exposures, the confidence intervals were wide and
included 1. There were, however, two striking findings. The
odds ratio for exposure to alkali was 3.5 (95% CI = 0.8-15.9),
which rose to 7.8 (95% CI = 1.7-35.3) when the analysis was
limited to self-respondents. Similarly, a history of exposure to
carbon monoxide was associated with an odds ratio of 2.9 (95%
J The abbreviation used is: CI, confidence interval.
CI = 1.0-8.4), which increased to 6.1 (95% CI = 2.0-18.2)
when only self-respondents were considered.
DISCUSSION
There are certain limitations in the data that must be taken
into account in interpreting this analysis. First, the analysis is
based upon only 69 cases. Many of the exposures examined
were rare. When exposed cases were few, the number of strati
fication variables had to be decreased to allow an estimate of
the relative risk. These estimates are associated with extremely
broad confidence intervals.
Second, our ascertainment of exposure status relied wholly
Table 4 Risk of light chain myeloma in relation to a history of radiation
exposure
% reported exposure
Cases
Exposure
All
Self
Controls
(n = 69) (n = 46) (n=1683)
Adjusted relative risk
All
respondents
Selfrespondents
RR° 95% CI RR 95% CI
X-rays>i/yr>l/l-5yrDental
Chest
X-rays>i/yr>l/l-5yr34.330.021.427.134.031.923.427.726.332.522.122.81.41.11.11.60.7-2.70.5
" RR. risk relative to that of persons exposed less often than once every 5
years, adjusted for age, sex, race, residence, and educational attainment.
4039
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1989 American Association for Cancer Research.
LIGHT CHAIN MYELOMA: CASE-CONTROL
Table .1 Risk of light chain myeloma
in relation
exposureCasesExposureVaccinationsSmallpoxTuberculosisTetanusPolioInfluenzaViral
% reporting
STUDY
to a history of infectious exposures
riskAll
relative
respondentsRR°1.11.00.80.91.00.90.50.50.70.72.01.00.00.00.01.10.60.91.10.50.20.49
(n46)87.210.651.146.855.363.883.053.26.434.02.12.10.00.00.027.76.453.22.18.54.32.1Controls
=
(n =1683)88.618.157.252.959.866.582.060.98.341.70.82.41.21.32.123.76.648.53.013.95.410.3Adjusted
69)82.912.947.141.454.354.368.644.34.331.41.41.40.00.00.027.14.345.74.37.12.95.7Self
=
CI0.5-2.50.5-2.30.5-1.50.5-1.90.6-1.60.5-1.70.3-1.10.3-1.00.2-2.10.4-1.3
CI0.3-2.90.3-1.90.4-1.40.4-1.80.5-1.7
illnessMumpsMeaslesOhickenpoxShinglesFever
blistersMononucleosis*HepatitisPolio'Bacterial
illnessOsteomyelitis'Tuberculosis0Urinary
infectionDivert
tract
iculosisTooth
abscessRheumatic
feverScarlet
feverVenereal
diseaseBronchitis''All(n
" RR. risk relative to that of persons with no history of that exposure, adjusted for age, sex. race, residence, and educational attainment.
'' Relative risk for all respondents and self-respondents adjusted for age. sex. and race.
1 Exact confidence intervals adjusted for age and race.
'' Relative risk for self-respondents adjusted for age. sex. and race.
on the recall of study subjects. Particularly for chemical expo
sures, for which we asked only whether an individual had ever
been "highly exposed," the potential for a great deal of misclassification is present. Certainly, the possibility exists that
persons with a disease are more likely to affirm an exposure
than are controls who are well. However, if a recall bias were
present, one might have expected many more positive findings
than actually were identified.
Third, our classification of chemical exposures for purposes
Table 5 Risk of light chain myeloma
exposureCasesExposureAcidsAldehydesAliphatic
of analysis required some judgment in determining what sub
stance was actually encountered by the subject. These decisions
were made without the knowledge of case-control status, and
so classification errors should have occurred at random.
Finally, given the large number of exposures examined in
this study, it is likely that a few positive findings occurred solely
by chance.
A history of a medical implant was positively associated with
the development of light chain myeloma. There was some
in relation to a history of chemical
% reported
riskAll
exposure
relative
respondentsRR°1.10.00.43.50.02.00.02.90.40.00.00.00.00.00.22.40.51.11.71.995%
CI*0.2-6.00.0-4.00.1-1.20.8-15.90.0-1.00.2-20.30.0-7.31.0-8.40.1-1.30.
(n 46)4.30.08.74.30.02.20.010.98.70.00.00.00.00.00.06.56.52.26.54.3Controls
=
(n =1683)2.51.214.41.84.30.60.73.414.03.50.70.20.40.10.04.316.91.84.81.5Adjusted
(n 69)2.90.05.82.90.01.40.07.27.70.00.00.00.00.00.05.87.21.47.22.9Self
=
CI0.3-8.60.0-6.50.2-2.21.7-35.30.0-1.5
hydrocarbonsAlkalisAromatic
hydrocarbonsAsbestosAsphyxiantsCarbon
monoxideChlorinated
hydrocarbonsDustsDyes
inksEstersEthersFertilizersOrganically
and
polymersMetalsOilsOther
high
causticsPaints
solventsPesticidesAll
and
' RR. risk relative to that of persons with no history ofthat exposure, adjusted for age. sex. race, residence, and educational attainment.
* Exact confidence interval, adjusted for age and race.
4040
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1989 American Association for Cancer Research.
LIGHT CHAIN MYELOMA: CASE-CONTROL
indication of a duration-response effect, which is what one
might expect if exposure to such foreign bodies were truly
causal. On the other hand, since light chain myeloma is a
disease that arises within the marrow space, it might be pre
dicted that orthopedic implants would be associated with a
particularly increased risk. This was not so. Among cases and
controls with a history of an implant, in a similar proportion
the implant was an orthopedic one. Furthermore, in the analysis
of all forms of multiple myeloma, a positive association between
a history of a medical implant and myeloma was not found (6).
A history of chickenpox rendered a moderate protective effect
of borderline significance. The effect was, however, limited to
subjects born between 1920 and 1930. If the effect in this cohort
is true, rather than the result of random variation, the expla
nation is obscure.
Alkali exposure was strongly associated with the development
of light chain myeloma, albeit with a wide confidence interval.
For all forms of multiple myeloma combined, however, the
relative risk associated with a history of alkali exposure was 1.2
(95% CI = 0.6-2.5) (2). While this difference between light
chain and other myeloma could reflect a true etiological differ
ence, the possibility of chance as an explanation must be borne
in mind.
The association between carbon monoxide exposure and light
chain myeloma was similarly strong and significant. While
carbon monoxide exposure was also a risk factor for all forms
of multiple myeloma combined (relative risk, 1.9; 95% CI =
1.1-3.2), with light chain myeloma cases removed from the
group, the odds ratio associated with carbon monoxide expo
sure fell to 1.3 (95% CI = 0.7-2.7).
The analysis of multiple myeloma, all types combined, also
produced positive associations for exposure to pesticides, paints
and solvents, and metals. Similar positive associations with
light chain myeloma could also exist but nonetheless have failed
to emerge in the current analysis, due to a lack of power to
detect real differences. Alternatively, these exposures may sim
ply not be risk factors for light chain myeloma.
The results of this study demonstrate several positive asso
ciations for light chain myeloma that were not seen for all forms
of myeloma combined. While these differences may have oc
curred by chance, they could be an indication that light chain
myeloma is, at least in part, etiologically distinct from the more
common forms of the disease. This hypothesis will require
further testing, a task that unfortunately will be difficult, given
the rarity of this form of multiple myeloma.
STUDY
APPENDIX
1: Substances included in exposure categories
Exposure category
Acids
Aldehydes and ketones
Aliphatic hydrocarbons
Alkalies
Aromatic hydrocarbons
Asbestos
Carbon monoxide
Chemical asphyxiants
Chlorinated hydrocarbons
Dusts
Dyes and inks
Esters
Ethers
Fertilizers
Metals
Oils
Organically high polymers
Other caustic substances
Paint and paint-related prod
ucts and/or other organic
solvents
Pesticides
Major substances included
Muriatic, hydrochloric, sull'urie, chromic,
nitric, and acetic acid; other miscella
neous acids
Formaldehyde, ketones, acetone
Gasoline, diesel, butane, propane, odivi
ene, acetylene, kerosene
Ammonia, sodium chlorate, caustic soda,
lime, cement
Benzene, toluene, xylene, creosote
Asbestos insulation, asbestos on brake
shoes, asbestos shipyard
Diesel, jet fuel, and automobile exhaust;
coal fumes; smoke
Hydrogen cyanide, hydrogen sulfide
Carbon tetrachloride, trichloroethylene,
méthylène
chloride
Cotton, textile, rock, wood, silica, coal,
plant, and animals dusts; Tiber glass
dust; insulation material dust
Dyes, beautician and/or hairdressing
products, inks
Nitroglycerine, phosphate esters
Ether
Fertilizers, nitrous soda fertilizer
Cast iron, lead vapors and liquid, arsenic,
gold, manganese, brass fumes, copper
powder and fumes, molybdenum
fumes, mercury vapors and liquid, tin,
silver, beryllium, cadmium
Oils, greases, grinding fluid
Plastic and rubber compounds
Chlorine, sulfur, fluorine, and nitrogen
compounds; phosgene; ozone
Paints, paint thinners and removers, lac
quers, lacquer thinners, varnish, var
nish removers, shellacs, glues, adhesives, other solvents
Pesticides, insecticides, organophosphorus and organochlorine compounds,
arsenical insecticides, pyrethrum, her
bicides, rodenticides
REFERENCES
1. Oken, M. M. Multiple myeloma. Med. Clin. N. Am., 68: 757-787, 1984.
2. Morris, P. D., Koepsell, T. D., Dating, J. R., et al. Toxic exposure and
multiple myeloma: a case-control study. J. Nati. Cancer Inst., 76: 987-994,
1986.
3. Mantel, N., and Haenszel, W. Statistical aspects of the analysis of data from
retrospective studies of disease. J. Nati. Cancer Inst., 22: 719-748, 1959.
4. Miettinen, O. S. Estimability and estimation in case-referent studies. Am. J.
Epidemici., 103: 226-235, 1976.
5. Schlesselman, J. J. Case Control Studies, p. 180. New York: Oxford Univer
sity Press, 1982.
6. Koepsell, T. D., Dating, J. R., and Weiss, N. S. Antigenic stimulation and
the occurrence of multiple myeloma. Am. J. Epidemici., Ì26:
1951-1962,
1987.
4041
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1989 American Association for Cancer Research.
Infectious and Noninfectious Exposures in the Etiology of Light
Chain Myeloma: A Case-Control Study
Ann R. Williams, Noel S. Weiss, Thomas D. Koepsell, et al.
Cancer Res 1989;49:4038-4041.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/49/14/4038
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 17, 2017. © 1989 American Association for Cancer Research.