Alberts, James W. MD, PhD; Wald, John J. MD; Werner, Robert A. MD; Franzblau, Alfred MD; Berent, Stanley PhD
On November 7 and 8, 1996, the Association of Occupational and Environmental Clinics and the National Institute for Occupational Safety and Health sponsored a conference in Washington, DC, on solvent exposure among railroad workers.1 At this conference, participants agreed that workers with solvent-associated occupational illnesses need to be defined more clearly than has been accomplished to date. It was recommended that workers with suspected solvent-induced neurotoxic disorders be reported so that the details about their disorders can be used to clarify the nature of solvent-associated illnesses and to better define the pathophysiology of identified disorders.
Several organic solvents, including n-hexane and methyl n-butyl ketone produce polyneuropathy after recreational2,3 or occupational4-9 exposures. The clinical features of this polyneuropathy include distal weakness, stocking-glove distribution sensory loss, and areflexia, with motor-conduction velocity slowing more than is typically associated with axonal-loss lesions. Sural nerve biopsy shows axonal degeneration with focal axonal swelling from accumulation of neurofibrils.3,6 The type and magnitude of clinical impairment is related to the potency, dose, and duration of exposure.10 In contrast, the association of clinically significant polyneuropathy after occupational exposure to other solvents, such as trichloroethylene, trichloroethane, perchloroethylene, mineral spirits, or similar solvents, alone or in combination, is controversial.4-9,11,12 Clinical descriptions are often uncontrolled, with some studies describing polyneuropathy and others showing no adverse effects, and supporting neuropathological and experimental animal studies are inconclusive or lacking.10,13-15 A distinction between clinical and subclinical polyneuropathy in these reports is not always clear, making interpretation and comparisons difficult.13
The mild neuropathies reported by Seppalainen and Antti-Poika in 60% of 87 patients diagnosed with chronic occupational solvent intoxication did not demonstrate a dose-response effect, and there was no evidence that termination of exposure influenced the abnormalities.16 The limitations of their study included uncertainty of the initial diagnosis, reliance upon conduction velocity without mention of response amplitude (a sensitive indicator of axonal loss lesions), and failure to control limb temperature. Orbaek et al reported reduced sural and median sensory amplitudes in 32 exposed subjects with solvent-induced toxic encephalopathy, compared with 50 unexposed control subjects, but quantitative vibration thresholds and motor nerve conduction study (NCS) results did not differ.17 In that study, the sensory NCS differences, described as small from a clinical point of view, were believed to reflect a mild axonal sensory polyneuropathy, but only two subjects had absent reflexes in the legs, clouding the distinction between clinical and subclinical polyneuropathy.17 Limb temperature and anthropometric features were not reported. Bleecker et al found significant correlations between increasing chronic low-level exposure to mixed solvents and increased vibration thresholds in workers from two paint-manufacturing plants.12 Their findings were subclinical, and typical symptoms of polyneuropathy were absent. The mild abnormalities consistent with polyneuropathy detected by Maizlish and associates in 16% of workers chronically exposed to mixed solvents disappeared when multiple linear regression models controlled for the effects of age, sex, alcohol intake, and examiner.11 Both of the latter studies utilized conventional clinical criteria and quantitative sensory testing for the detection of polyneuropathy, raising doubt about the presence of clinically significant polyneuropathy in association with mixed solvent exposure.
The peripheral and central nervous systems are potential targets for solvent-induced neurotoxicity. It is important to establish whether the peripheral nervous system is involved in cases of solvent-induced encephalopathy, in part because establishing the presence of polyneuropathy would reduce the controversy surrounding solvent-induced encephalopathy by providing additional evidence of nervous system involvement, making the diagnosis of a neurotoxic disorder more readily apparent. The diagnosis of polyneuropathy is relatively objective and is not influenced by the numerous factors such as education, anxiety, and motivation that make examination of the central nervous system difficult. We therefore proposed to determine whether objective clinical or subclinical evidence of polyneuropathy existed in a population of railroad workers, and we reviewed neurologic and electrodiagnostic assessments of 30 railroad workers evaluated by us in the context of litigation related to solvent neurotoxicity. All of the workers had previously been diagnosed with toxic encephalopathy by either a neurologist or an occupational medicine physician, and the same physicians believed that several of the workers had toxic polyneuropathy. For most subjects, the diagnostic criteria developed at the 1985 World Health Organization (WHO) workshop on organic solvents had been used by these clinicians to establish the presence of toxic encephalopathy.13,18 This 1985 scheme includes a gradation of severity, with the mildest form of encephalopathy (type 1) based on symptoms that were not necessarily specific, although they were usually referable to the central nervous system. Mild toxic encephalopathy is defined in two ways, depending on the presence of sustained change in mood or personality change (type 2A) or an intellectual impairment on neuropsychological testing (type 2B). Subjects with severe toxic encephalopathy (type 3) are described as having features of chronic dementia. Based on reports of previous examiners, these subjects were representative of railroad workers with solvent-induced neurotoxicity. We report the peripheral nervous system evaluations of these same railroad workers who were believed to have had substantial occupational exposure to mixed solvents.
All subjects were employed in the railroad industry and involved in litigation against their employer because of alleged occupational solvent exposure producing neurotoxicity. Defense attorneys referred the subjects to us for independent medical examinations. All subjects knew that their examinations were performed in the context of their litigation. Data from a randomly chosen subsample of eligible subjects were included in the analyses. All 30 selected subjects had a previous diagnosis of toxic encephalopathy; these diagnoses had usually been made on the basis of the WHO classification of encephalopathy.18 For subjects whose type of encephalopathy had not been specified by their physician, we applied the WHO terminology. The distribution of subjects on the basis of encephalopathy type was as follows: type 1 (n = 6), type 2A (n = 15); type 2B (n = 3); type 2A, B (coexisting type 2A and 2B; n = 5), and type 3 (n = 1).
The characteristics of the subjects are listed in Table 1. Exposure histories were subjective and based on the individual's description. Despite a variety of individual jobs providing different exposure opportunities, most subjects reported similar exposure experiences because they had worked together in common environments. For example, an electrician might experience exposure to solvents being used by a mechanic working on the same locomotive. A typical history included daily spraying of up to several 55-gallon drums of solvent from a pressurized tank onto the locomotive, as well as direct application with a rag from an open "solvent bucket." Respiratory or dermal protection was not typically used, and several subjects reported washing their work clothes in the solvents to remove grease. Subjects typically referred to the chemicals used by common or trade names. On the basis of this information, we determined that the solvents most frequently used were trichloroethylene, trichloroethane, and perchloroethylene, either alone or in combination. Other solvents included carbon tetrachloride, xylol, mineral spirits, kerosene, and toluene, as well as unidentified solvent mixtures. The mean duration of occupational solvent exposure was 20 years, and all subjects reported exposures exceeding 10 years. These exposure durations fulfilled the criteria used by Edling et al19 for long-term exposure. Fourteen workers reported ongoing (current) solvent exposure, although it was typically at a lower level than that experienced in previous years because of changing work practices. The highest-level solvent exposures occurred prior to the mid-1980s for most subjects. All subjects reported acute solvent intoxication on a regular basis, symptoms of which included headache, dizziness, "drunkenness," and balance problems. No subject reported accidents, injuries, or loss of consciousness related to acute solvent intoxication.
Two subjects had adult-onset diabetes mellitus for many years, with elevated glucose levels that had not required insulin treatment. Three subjects reported prior alcohol abuse (remote past); one of the three reported prior polysubstance abuse. No subject was taking medications known or suspected to have peripheral neurotoxicity. Several subjects were being treated with tricyclic antidepressants, which are sometimes associated with distal paresthesias, or antihypertensive medications that are occasionally associated with symptoms of dysautonomia. Four subjects had known carpal tunnel syndrome (three bilateral), and three of them reported previous carpal tunnel surgery.
All subjects underwent standard clinical neurologic examinations.11 The medical history included an occupational history and a review of past and current use of medications, alcohol, and chemicals used in hobbies. The history included items that were potentially related to polyneuropathy, such as the distribution and characterization of abnormal sensation (numbness, tingling, and pain) and strength. Medical records were reviewed to identify confounding medical problems unrelated to solvent exposure that could adversely affect the nervous system. Previous neurologic examinations and test results were also reviewed. The neurologic examination included an evaluation of cranial nerves; motor function, including station and gait, coordination, alternate motion rate, and strength; subjective sensory function, including fine touch, vibration (128 Hz), pin-pain, temperature, dual simultaneous stimulation, and Romberg test; and reflexes. Strength was evaluated in proximal and distal muscles, including the hand and foot intrinsic muscles. Muscle-stretch reflexes were graded on a five-point scale. Provocative tests, such as orthostatic blood pressure measurements and volitional hyperventilation, were evaluated when indicated.
Nerve Conduction Studies
Sensory and motor NCSs were performed on the dominant side, unless the other side was more symptomatic, in which case that side was examined (four subjects). A basic study included an evaluation of sural, peroneal motor, median sensory and motor, and ulnar sensory nerves. Some subjects had additional nerves evaluated to address specific complaints or findings. Standard techniques of supramaximal percutaneous nerve stimulation and surface recording were used, based on anatomic landmarks and standardized stimulation to recording electrode distances.20,21 Sensory nerve action potential (SNAP) and compound muscle action potential amplitudes were measured from the baseline to the negative peak and reported after stimulation at the distal site. Conduction velocity was calculated using the response onset for the forearm and leg segments, and a terminal conduction velocity was calculated by dividing the distal distance by the terminal latency to the SNAP onset. F-response latencies were measured as the minimal latency in a series of 10 to 15 F responses, using distal antidromic stimulation. Skin temperatures were monitored at the palm and at the calf between stimulating and recording electrodes, using an analog thermometer. Temperature was monitored and limbs warmed above 31°C, using moist heat placed over the forearm or calf if necessary. The mean palm temperature was 33.7°C (range, 32.0 to 34.8°C); the mean calf temperature was 33.2°C (range, 31.3 to 34.9°C).
Assessment of Polyneuropathy
The diagnosis of subclinical or clinical polyneuropathy was established using a combination of abnormalities from the categories of symptoms, signs, and electrodiagnostic testing that is consistent with standard clinical practice.22,23 To be considered positive, symptoms and signs had to be judged by the examining neurologist as consistent with a symmetric sensory or sensorimotor polyneuropathy involving the lower or upper and lower extremities. For the purposes of this study, a liberal interpretation was applied and even transient symptoms were included. Clinical polyneuropathy was defined as an abnormal clinical examination consistent with a symmetric sensory or sensorimotor polyneuropathy. This required, at minimum, abnormality in at least two of the following categories: symptoms, decreased sensation consistent with large- or small-fiber loss, or decreased or absent ankle reflexes. Confirmed clinical polyneuropathy required electrodiagnostic confirmation, consisting of abnormal NCS results in two or more distinct anatomic sites on the basis of evaluation of the sural, peroneal motor, and median sensory and motor nerves.24,25 Absolute NCS abnormality of any individual measurement was determined on the basis of published normal data.26-28 Subclinical polyneuropathy was defined as symptoms or electrodiagnostic abnormality without appropriate clinical signs. Small-fiber polyneuropathy was defined clinically as symptoms of dysautonomia (postural lightheadedness, impotence, gastroparesis) or painful dysesthesias plus signs of symmetrical stocking-glove decreased pin-pain or temperature sensations, or orthostatic hypotension (supine to standing systolic blood pressure drop of at least 10 mm Hg without increase in pulse rate).
In addition to creating a tabular report of the clinical findings, we compared our electrodiagnostic results with those of several historical control groups. NCS results were compared with (1) University of Michigan adult normal values,29,30 (2) consensus lower and upper limits of normal provided by electromyographers in a past pharmacological investigation that used similar techniques,26 (3) US Centers for Disease Control (CDC) data,31 (4) results previously obtained from active workers who were asymptomatic for hand complaints,32 and (5) a control group established from a population of active workers without occupational exposure to solvents or other chemicals.28 We established the latter control group by matching each subject with three controls (or a minimum of two controls if a third matched control was unavailable) selected by gender, age (±3 years), and body mass index (BMI; ±2 kg/m2).33,34 We had 80% power to detect approximately a 5-µV difference in the median and a 6-µV difference in the ulnar sensory mean amplitudes in the comparisons of the railroad workers with the matched control group, with an alpha of 0.05 (two-sided).35 This calculation was based on log transformation of the amplitude data according to routine statistical practice to make the distribution more symmetrical. Because many NCS measures do not have a symmetrical distribution, group comparisons were made using the nonparametric Mann-Whitney U test. Because multiple NCS measures were compared for each analysis, a P value of 0.01 was used to determine statistical significance (approximate Bonferroni adjustment).36
NCS results were compared for groups of railroad workers classified according to the presence of polyneuropathy symptoms, previous diagnosis of polyneuropathy, and disability status. We also compared NCS results for groups classified by the severity and type of encephalopathy. For this comparison, we assumed that type 1 encephalopathy was less severe than type 2 and performed a dichotomous comparison of NCS results (t test). Potential indicators of dose-response relationships were evaluated using simple linear and stepwise regression models in which occupational solvent exposure duration, job title, age, height, weight, BMI, and temperature were used as independent variables, and nerve conduction measures were the dependent variables. For the NCS comparisons of groups separated by the type of encephalopathy, we included the single subject with type 3 encephalopathy into the type 2 A, B group before evaluating for group differences (analysis of variance).
The study protocol was reviewed and approved by the University of Michigan Medical School Institutional Review Board for Human Subject Research (IRBMED).
The presence of symptoms, signs, and electrodiagnostic abnormalities consistent with polyneuropathy is summarized in Table 2 for the group of 30 subjects. Eight subjects had symptoms that were potentially related to polyneuropathy, consisting primarily of bilateral sensory loss, numbness, or tingling. Three of the eight subjects with potential symptoms of polyneuropathy and three additional subjects had previously been diagnosed with toxic polyneuropathy. At the time of our evaluation, only three of the six subjects previously diagnosed with toxic polyneuropathy reported distal symmetric numbness or tingling. Another subject with distal foot numbness was taking a tricyclic antidepressant that has been associated with extremity numbness, tingling, and paresthesias. Ten subjects had symptoms potentially associated with autonomic dysfunction, but only one (a subject with diabetes mellitus) demonstrated an abnormal orthostatic blood pressure drop upon standing. Four of the 30 subjects had subjective evidence of distal vibration loss on clinical examination, and one of the four also had diminished pin-pain sensation in a stocking distribution. Three of the four reported potential symptoms of polyneuropathy, and two of the three were previously diagnosed with toxic polyneuropathy. Ankle reflexes were normal (2+) in all but one subject, who had absent ankle reflexes. This symptomatic subject was one of the two subjects with diabetes mellitus. The other diabetic subject was the only worker who demonstrated orthostatic hypotension.
Ten of the 30 subjects had at least one NCS abnormality among the ten measures evaluated for the four test nerves. Of these ten subjects, one had multiple abnormalities, including reduced sural amplitude (generally considered the "gold standard" for polyneuropathy) and prolonged distal latency. Two subjects fulfilled absolute and relative criteria for a median mononeuropathy at the wrist (carpal tunnel syndrome), and three subjects had other median nerve abnormalities not localized to the wrist. Four subjects had isolated peroneal abnormalities, including reduced conduction velocity (1), reduced amplitude consistent with local trauma (1), and prolonged F-wave latencies not corrected for height (3). Twenty-seven subjects had ulnar sensory NCSs performed. All 27 had normal amplitude responses, but two of the 27 had slightly prolonged distal latencies. The three subjects who did not have ulnar recordings had no abnormalities on remaining NCSs.
The classification of polyneuropathy is listed in Table 3. Four subjects fulfilled the minimal criteria for clinical polyneuropathy, with symptoms and findings consistent with a mild sensory polyneuropathy in all four, but only one of the four had confirmed clinical neuropathy on the basis of NCS results. Of the four, one had diabetes mellitus and one had a history of prior alcohol abuse. The subject with diabetes mellitus also had abnormal ankle reflexes, and when NCS results were included, he was the only subject who fulfilled the criteria for confirmed clinical polyneuropathy. Nine subjects reported postural lightheadedness or impotence, symptoms potentially related to a small-fiber or autonomic neuropathy (Table 2). Only one subject had clinical signs of small-fiber polyneuropathy, demonstrating orthostatic hypotension. This subject had diabetes mellitus, a common cause of autonomic neuropathy, and he was also taking a diuretic medication. Of the remaining eight, another had diabetes mellitus and six were using medications (including alprazolam, diltiazem, fluoxetine, hydrochlorothiazide, paroxetine, sertraline, and trazodone) that have frequently been associated with symptoms suggestive of dysautonomia.
The NCS results are summarized in Table 4, showing comparison to several historical control groups. Few individual measures exceed the lower or upper limits of conventional normal values, including those used in our laboratory.26,29 In terms of average values for the group, only the median and ulnar sensory studies demonstrated substantial mean differences (21 v 38 µV, and 21 v 32 µV, respectively), with few other differences of clinical importance.29,30 Sural responses, generally considered to be the most sensitive measures of a distal axonal polyneuropathy, were similar in both groups (amplitude, 14 v 17 µV; conduction velocity, 47 v 48 m/s, respectively). Comparison with other published control values, such as the 4462 randomly selected male veterans studied by the CDC in a search for adverse health effects among men who had served in Vietnam,31 identified few group differences. Similarly, comparisons of the railroad workers with the group of 137 neurologically intact active industrial workers without exposure to solvents or other chemicals32 demonstrated similar average median and ulnar sensory measures (21 v 24 µV and 21 v 23 µV, respectively). In this comparison, the mean median motor amplitude is lower in the railroad worker group, compared with the industrial worker control group, but is comparable with the other control groups. Because median and ulnar sensory response measures are sensitive to a variety of factors,32 we developed a control group of active workers without solvent exposure, matched by gender, age, and BMI to the railroad workers.33,34 Average median and ulnar sensory amplitudes and terminal conduction velocities for the railroad workers did not differ significantly from those for the matched industrial control group.
Subdivision of the 30 railroad workers into groups selected by the presence or absence of symptoms potentially related to polyneuropathy, previous diagnosis of polyneuropathy, and disability status demonstrated no significant differences between groups (Table 5). Similarly, comparison of electrodiagnostic measures for subjects in the sample presented here who were disabled and not working with workers who were actively employed demonstrated no significant differences (Table 5). We also found no significant group differences related to age, height, weight, BMI, or occupational solvent exposure duration. Neither duration of exposure nor job title significantly influenced NCS measures in multiple linear regression models. Nerve conduction measures did not differ significantly between groups after workers were separated on the basis of their classification of encephalopathy. This included a comparison of workers with type 1 and type 2 encephalopathy and additional comparisons after further subdivision into groups identified as having types 1, 2A only, 2B only, and coexisting 2A and 2B (2 A, B) encephalopathy.
The 30 railroad workers we report on in this article had been diagnosed by other health care providers as having solvent-associated toxic encephalopathy, but only four fulfilled the criteria for clinical polyneuropathy. Three of the four subjects fulfilled symptom and sensory loss criteria only, and all three had normal ankle reflexes. The single subject who had confirmed clinical polyneuropathy was the only subject found by us or any previous examiner to have absent ankle reflexes, an objective indication of polyneuropathy. This subject also was one of two subjects in the group with a systemic illness known to cause polyneuropathy (diabetes mellitus). The other subject with diabetes mellitus had clinical evidence of mild dysautonomia, characterized by impotence, probable gastroparesis, positional lightheadedness, and postural hypotension. The NCS criteria used to confirm clinical polyneuropathy or identify subclinical polyneuropathy are conventional and have been utilized by others.24,26 According to these criteria, focal lesions of individual nerves are excluded by a requirement that there be an abnormality in at least two anatomically separated locations. The only worker with confirmed polyneuropathy had diabetes mellitus, supporting the sensitivity of the criteria. This subject had an elevated median-to-sural SNAP amplitude ratio of almost 7 (normal, 1.9; range, 0.9 to 4.0), typical of diabetic polyneuropathy.37 The other diabetic subject had evidence of dysautonomia but no evidence of a sensory or sensorimotor polyneuropathy.
NCSs are frequently used to identify subclinical polyneuropathy. Electromyographers recognize the importance of limb temperature and subject age when making group comparisons. However, only recently has the influence of anthropometric measures such as height and BMI been recognized as important in the establishment of normal values.28,31,38-40 Failure to consider these potential confounders increases the likelihood of attributing minor deviations from normal to polyneuropathy. Comparison of mean NCS results obtained from the railroad workers to generic control values suggested substantial reductions in median and ulnar sensory amplitudes, and comparison to an industrial worker control group suggested the lower median motor amplitudes as well. This latter finding did not appear to be of direct clinical importance, and it likely reflected an unusually high amplitude for this industrial worker control group, compared with the other control groups, not an isolated reduction in the railroad worker group. The median motor amplitude measure is subject to substantial variability, and it is not particularly sensitive to generalized polyneuropathy, compared with lower extremity motor responses or sensory responses. Comparison of the railroad worker NCS results, exclusive of the two diabetic workers, with CDC data demonstrated significant slowing of median terminal conduction velocity in the railroad worker group, which likely reflects the inclusion of workers with median mononeuropathy at the wrist. Nevertheless, all sensory differences disappeared in comparisons to active occupational control workers matched by gender, age, and BMI. In addition, neither duration of exposure nor job title significantly influenced NCS measures in multiple linear regression models, there being no evidence of subclinical dose-response relationships. Failure to identify NCS differences among groups of subjects separated by encephalopathy type further argues against a dose-response relationship and makes the presence of a subclinical dose-response effect on NCS measures unlikely.
It is surprising that more differences between the railroad workers and the control groups were not identified, because the criteria used to select control subjects would have excluded several railroad workers independent of occupational solvent exposure. For example, reasons for exclusion from the control groups typically included any diseases or exposures associated with polyneuropathy, including the presence of diabetes mellitus or a history of current or previous alcohol abuse. To put the railroad worker NCS data presented in this study into perspective and demonstrate the sensitivity of the NCS measures, the NCS results from this study can be compared with those reported for male patients with type 1 diabetes mellitus who fulfill the criteria for mild diabetic polyneuropathy.41 Significant differences are apparent for peroneal and sural measures, and mean amplitudes and conduction velocities for diabetic patients are close to the lowest values reported for the railroad workers, exclusive of the two diabetic workers. For example, the diabetic subjects had a mean peroneal amplitude of 3.2 mV, approximately 50% of the mean railroad worker amplitude, and a conduction velocity of 39 m/s, more than 5 m/s slower than the railroad worker mean value. Sural differences are even more impressive, with mean amplitudes of 5 v 15 µV and mean conduction velocities of 39 v 47 m/s for diabetic subjects and railroad workers, respectively. These comparisons indicate that it is very unlikely that a subclinical polyneuropathy is being overlooked in this group of railroad workers.
The existence of polyneuropathy in patients diagnosed with solvent-induced toxic encephalopathy should not be controversial, unless it is particularly mild or exceedingly rare. The peripheral nervous system is easily evaluated, and peripheral measures are unaffected by educational, emotional, and motivational influences. Failure to identify unexplained polyneuropathy in solvent-exposed workers diagnosed by others to have toxic encephalopathy must be reconciled with previous studies associating solvent-induced encephalopathy and polyneuropathy.16,17 One possibility is uncertainty about the diagnosis of toxic encephalopathy, not only in previous studies but in this study as well. The lack of uniform diagnostic criteria leaves uncertainty about the diagnosis of encephalopathy, and evidence of polyneuropathy would have provided support for a toxic etiology. Another consideration is the limited exposure information. The reported exposure durations of 10 to 29 years exceeded those described as sufficient to produce encephalopathy.15 The magnitude of exposure is unknown, but all workers reported acute intoxication on a regular basis. Furthermore, a participant at the 1996 Conference on Solvent Exposure Among Railroad Workers described similar railroad workers as possibly representing the largest cohort of workers in the United States with a history of such heavy solvent exposure.42 There was nothing to indicate that the workers we describe were atypical of their peers. Nevertheless, monitoring was not utilized, and only approximations of exposure are available, based on the workers' descriptions. The absence of unexplained polyneuropathy suggests that symptoms such as numbness, tingling, weakness, or imbalance are unlikely to represent peripheral nervous system dysfunction in the railroad workers claiming neurotoxic injury from occupational solvent exposure.
If these 30 railroad workers do not have polyneuropathy, how do we explain the symptoms and signs suggestive of polyneuropathy that have been described by other examiners? One possibility involves the misinterpretation of nonspecific symptoms as representing polyneuropathy. The presence of symptoms suggestive of polyneuropathy in 8 of 30 subjects (27%) seems excessive, although similar symptoms in combination with mild signs of polyneuropathy were reported previously in 16% of workers with occupational solvent exposure.11 In that study, however, evidence of polyneuropathy was independent of exposure. Workers with the lowest measured solvent exposure had more peripheral symptoms, on average, than did workers in the highest exposure group. Furthermore, the mean number of peripheral symptoms was similar for workers classified as exposed, compared with those classified as unexposed.11 This suggests that symptoms believed to be suggestive of peripheral nervous system dysfunction are nonspecific and relatively prevalent in the general population without polyneuropathy.
In addition, many symptoms reported by the subjects we evaluated had explanations that were unrelated to peripheral nervous system dysfunction. Concern among these workers about the adverse health effects of solvent exposure is likely associated with increased vigilance and awareness of symptoms that previously would have gone unreported. Many symptoms in the subjects we studied were fleeting and inconsistent with polyneuropathy: symptoms of polyneuropathy vary in intensity, but typically do not appear and disappear. Some symptoms were reproduced by volitional hyperventilation, suggesting that they were anxiety-related and not associated with polyneuropathy. Numerous medications, including tricyclic antidepressants, produce numbness, tingling, and paresthesias, although tricyclic antidepressants are frequently prescribed to treat dysesthesias. Other symptoms were asymmetric or focal and were consistent with localized problems such as median mononeuropathy at the wrist or digital nerve compression, not diffuse polyneuropathy. The absence of significant NCS differences between workers with and without sensory symptoms indicates that the symptoms were not related to polyneuropathy. Reliance on subjective sensory findings with disregard for the magnitude, distribution, or modality of sensory loss causes further diagnostic confusion. Most toxic neuropathies involve large axons, producing distal vibration sensation and reflex abnormalities. Non-physiologic findings, such as an abrupt demarcation of abnormal to normal sensation, are inconsistent with a peripheral polyneuropathy, as are preserved ankle reflexes or a normal Romberg sign in the presence of markedly abnormal vibration sensation. Similarly, decreased sensation over heavily callused areas or in a distribution commonly associated with local trauma (eg, digital nerves in the foot) should not be confused with diffuse polyneuropathy. Even anthropomorphic features must be considered. Dyck and associates recently reported decreased sensation with increasing BMI in randomly selected subjects without known polyneuropathy, perhaps reflecting the distribution of a fixed number of sensory receptors over a larger surface area.23 This explanation applies to several of the reported railroad workers believed to have abnormal sensation. We were sensitive to the diagnostic pitfalls described above that are associated with polyneuropathy. We utilized conventional criteria to identify polyneuropathy and required NCS confirmation, the "gold standard" for diagnosing polyneuropathy.
We also considered the possibility these 30 workers experienced resolution of a previous toxic polyneuropathy. There are descriptions of recovery to normal or almost normal for patients with occupational-related polyneuropathy,43 although other researchers report an increased frequency of neuropathic abnormalities upon re-examination many years after workers had been removed from solvent exposure.16 None of the workers we report described resolution of previous symptoms suggestive of polyneuropathy nor had any sought medical evaluation that had resulted in a diagnosis of polyneuropathy (other than 1 one of the two workers with diabetes mellitus).
In closing, the study presented here found no evidence of confirmed polyneuropathy in 29 of 30 railroad workers with histories of substantial occupational exposure to mixed solvents and alleged toxic encephalopathy. The only subject with confirmed polyneuropathy had diabetes mellitus, one of the most common causes of polyneuropathy. There was also no evidence of a polyneuropathy that had improved after removal from exposure nor subclinical evidence of a dose-response solvent effect on NCS measures. Although these findings do not exclude the possibility of solvent-induced polyneuropathy in selected occupational situations, we did not find any evidence of a polyneuropathy in these railroad workers who had been selected by others as having characteristic solvent-induced toxic encephalopathy.
Partial funding for this study was provided by CSX Transportation, Inc. (contract sponsor) and a SPHERE (Supporting Public Health and Environmental Research Efforts) Award from the Dow Chemical Company Foundation.
1. Proceedings from the Workshop on Solvent Exposure Among Railroad Workers; organized by the Association of Occupational Environmental Clinics and sponsored by the National Institute for Occupational Safety and Health; November 7-8, 1996; Washington, DC.
2. Kuwabara S, Nakajima M, Tsuboi Y, Hirayama K. Multifocal conduction block in n-hexane neuropathy. Muscle Nerve
3. Smith AG, Albers JW. n-Hexane neuropathy due to rubber cement sniffing. Muscle Nerve
4. Chang CM, Yu CW, Fong KY, et al. n-Hexane neuropathy in offset printers. J Neurol Neurosurg Psychiatry
5. Scelsi R, Poggi P, Fera L, Gonella G. Industrial neuropathy due to n-hexane: clinical and morphological findings in three cases. Clin Toxicol.
6. Herskowitz A, Ishii N, Schaumburg HH. n-Hexane neuropathy. a syndrome occurring as a result of industrial exposure. N Engl J Med.
7. Paulson GW, Waylonis GW. Polyneuropathy due to n-hexane. Arch Intern Med.
8. Iida M. Neurophysiological studies in n-hexane polyneuropathy in the sandal factory. Electroencephalogr Clin Neurophysiol.
9. Spencer PS, Schaumburg HH, Raleigh RL, Terhaar CJ. Nervous system degeneration produced by the industrial solvent methyl n-butyl ketone. Arch Neurol.
10. Spencer PS, Schaumburg HH. Organic solvent neurotoxicity: facts and research needs. Scand J Work Environ Health.
11. Maizlish NA, Fine LJ, Albers JW, Whitehead L, Langolf GD. A neurological evaluation of workers exposed to mixtures of organic solvents. Br J Ind Med.
12. Bleecker ML, Bolla KI, Agnew J, Schwartz BS. Dose-related subclinical neurobehavioral effects of chronic exposure to low levels of organic solvents. Am J Ind Med.
13. Baker EL, Fine LJ. Solvent neurotoxicity: the current evidence. J Occup Med.
14. Seppalainen AM. Neurotoxicity of organic solvents in occupational exposure. Scand J Work Environ Health
15. Baker EL. A review of recent research on health effects of human occupational exposure to organic solvents: a critical review. J Occup Med.
16. Seppalainen AM, Antti-Poika M. Time course of electrophysiological findings for patients with solvent poisoning: a descriptive study. Scand J Work Environ Health.
17. Orbaek P, Rosen I, Svensson K. Electroneurographic findings in patients with solvent induced central nervous system dysfunction. Br J Ind Med.
18. World Health Organization. Chronic Effects of Organic Solvents on the Central Nervous System and Diagnostic Criteria.
Copenhagen: World Health Organization and Nordic Council of Ministers; 1985.
19. Edling C, Ekberg K. Ahlborg Jr G, Alexandersson R. Long term follow up of workers exposed to solvent. Br J Ind Med.
20. Evans BA. Nerve action potentials. In: Daube JR, ed. Clinical Neurophysiology.
Philadelphia: F.A. Davis Co.; 1996:147-156.
21. Daube JR. Compound muscle action potentials. In: Daube JR, ed. Clinical Neurophysiology.
Philadelphia: F.A. Davis Co.; 1996:199-234.
22. Dyck PJ. Detection, characterization, and staging of polyneuropathy: assessed in diabetes. Muscle Nerve.
23. Dyck PJ, Litchy WJ, Lehman KA, Hokanson JL, Low PA, O'Brien PC. Variables influencing neuropathic endpoints: the Rochester Diabetic Neuropathy Study of Healthy Subjects. Neurology.
24. The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med.
25. Robinson LR, Micklesen BS, Wang L. Strategies for analyzing nerve conduction data: superiority of a summary index over single tests. Muscle Nerve.
26. The Diabetes Control and Complications Trial Research Group. Effect of intensive diabetes treatment on nerve conduction in the Diabetes Control and Complications Trial. Ann Neurol.
27. O'Brien PC, Dyck PJ. Procedures for setting normal values. Neurology.
28. Stetson DS, Albers JW, Silverstein BA, Wolfe RA. Effects of age, sex, and anthropometric factors on nerve conduction measures. Muscle Nerve.
29. Albers JW, Cavender GF, Levine SP, Langolf GD. Asymptomatic sensorimotor polyneuropathy in workers exposed to elemental mercury. Neurology.
30. Greene DA, Sima AAF, Albers JW, Pfeifer MA. Diabetic neuropathy. In: Rifkin H, Porte D Jr, eds. Ellenberg and Rifkin's Diabetes Mellitus: Theory and Practice.
New York: Elsevier; 1990:710-755.
31. Letz R, Gerr F. Covariates of human peripheral nerve function: I-nerve conduction velocity and amplitude. Neurotoxicol Teratol.
32. Stetson DS, Silverstein BA, Keyserling WM, Wolfe RA, Albers JW. Median sensory distal amplitude and latency: comparisons between nonexposed managerial/professional employees and industrial workers. Am J Ind Med.
33. Werner RA, Franzblau A. Hand dominance effect on median and ulnar sensory evoked amplitude and latency in asymptomatic workers. Arch Phys Med Rehabil.
34. Franzblau A, Werner RA, Valle J, Johnston E. Workplace surveillance for carpal tunnel syndrome: a comparison of methods. J Occup Rehabil.
35. Freiman JA, Chalmers TC, Smith H Jr, Kuebler RR. The importance of beta, the type II error and sample size in the design and interpretation of the randomized control trial. N Engl J Med.
36. O'Brien PC, Shampo MA. Statistical considerations for performing multiple tests in a single experiment: 2-comparisons among several therapies. Mayo Clin Proc.
37. Bromberg MB, Albers JW. Patterns of sensory nerve conduction abnormalities in demyelinating and axonal peripheral nerve disorders. Muscle Nerve.
38. Werner RA, Albers JW, Franzblau A, Armstrong TJ. The relationship between body mass index and the diagnosis of carpal tunnel syndrome. Muscle Nerve.
39. Stetson DS. Median and Ulnar Conduction Measures in Control and Industrial Populations: Associations with Ergonomic Risk Factors
[doctoral thesis]. Ann Arbor: University of Michigan; 1991:1-251.
40. Rivner MH, Swift TR, Crout BO, Rhodes KP. Toward more rational nerve conduction interpretations: the effect of height. Muscle Nerve.
41. Albers JW, Brown MB, Sima AAF, Greene DA, for the Tolrestat Study Group for the Early Diabetic Intervention Trial. Nerve conduction measures in mild diabetic neuropathy: the effects of age, sex, type of diabetes, disease duration, and anthropometric factors. Neurology.
42. Frumkin H, Ducatman A, Kirkland K. Solvent exposure in the railroad industry (letter). J Occup Environ Med.
43. Allen N, Mendell JR, Billmaier DJ, Fontaine RE, O'Neill J. Toxic polyneuropathy due to methyl n-butyl ketone: an industrial outbreak. Arch Neurol.
© 1999 Lippincott Williams & Wilkins, Inc.