Correspondence DO PATIENTS "lITH OBSTRUCTIVE SLEEP APNEA HAVE THICK NECKS?
To the Editor: I read with interest the controversy between Partinen and Hoffstein (1)regarding the relative importance of the body mass index (BMI) and neck circumference in accounting for individual differences in the apnea/hypopnea index (AHl). The real issue underlying use of multiple linear regression is not which variable is entered first or second in the regression but, rather, if the assumptions of the regression are being met. To the extent that many of the key variables entered in the regression were highly correlated with each other (table 2 of the paper [2]), the assumption of absence of colinearity was violated, and any regression performed on these data will be invalid unless such redundancy is first removed. When the 10variables shown in table 2 were reduced by entering the correlation coefficients into a principal components analysis with varimax rotation, four factors emerged accounting for 76.1070 of the cumulative variance. Factor loadings are shown below.
AHI Age BMI EXT LEN PRO MID DIS GLO TRA 070 var
Fac I
Fac II
Fac III
Fac IV
- .15 .07 - .11 -.01 -.10 -.05 .28 .80* .94* .90* (31.2)
.74* .05 .87* .89* .10 -.03 - .19 - .18 - .18 .03 (18.3)
-.08 -.05
.30 .87* - .11 .02 ,42 - .25 .07 .03 - .01 .03 (10.2)
.00 -.06 .57* .74* .87* ,41
.04 - .16 (16,4)
* Primary factor loading of each variable. Note that BMI and external neck circumference (EXT) load on the same factor. This means that within these data there is no way to distinguish the relative importance of BMI and external neck size. The overlap within the data simply will not allow for it. The advent and availability of powerful statistical packages have allowed marvelously rich and complex data sets to be analyzed with relative ease. Unfortunately whether the assumptions underlying use of statistical procedures are met is still very much left to the individual investigator and, as such, must be examined carefully. DONALD
L.
BLIWISE, PH.D.
Stanford University Medical Center Stanford, California 1. Partinen M, Hoffstein V. Letters to the editor. Am Rev Respir Dis 1991;143:204.
2. Katz I, StradlingJ, SlutskyAS, ZamelN, Hoffstein V. Do patients with obstructive sleep apnea have thick necks? Am Rev Respir Dis 1990; 141: 1228-31.
VIRUS-INDUCED AIRWAY HYPERRESPONSIVENESS - POSSffiLE INVOLVEMENT OF NEURAL MECHANISMS
To the Editor: We enjoyed reading Dr. Laitinen's article that demonstrated airway hyperresponsiveness to histamine after infection with live attenuated influenza virus in previously normal subjects (1). Because histamine both stimulates vagally induced reflex bronchoconstriction (2) and releases tachykinins from sensory nerves (3), we suggest that the described virus-induced hyperresponsiveness may be neurally 1422
mediated. Indeed, the same group has shown that hyperresponsiveness to histamine in patients with naturally occurring viral infections can be blocked by atropine (4). Likewise, virus-induced hyperresponsiveness to histamine can be blocked by vagotomy in the guinea pig (5). These observations are consistent with our recent finding that parainfluenza virus infection in guinea pigs selectively damages inhibitory muscarinic receptors on pulmonary parasympathetic nerves (6). These receptors (of the M 2 subtype) normally provide a negative feedback whereby acetylcholine released from the vagus nerve turns off further release of acetylcholine (7). When these receptors do not function (as is the case in virus-infected guinea pigs), the bronchoconstriction caused by electrical stimulation of the vagus is markedly increased and cannot be suppressed by a M 2 agonist such as pilocarpine. This loss of function may be caused by viral neuraminidase (an enzyme present in both influenza and parainfluenza viruses) cleaving off sialic acid residues from the M 2 receptor. We have demonstrated, using ligand binding studies, that agonist affinity for the M 2 receptor is markedly decreased when the receptor is exposed to parainfluenza virus in vitro (8). This effect can be mimicked by an equivalent amount of Clostridium perfringens neuraminidase and blocked by 2,3-dehydro-2-deoxy-N-acetylneuraminic acid, a neuraminidase inhibitor. Alternatively, virus-induced loss of airway neutral endopeptidase activity (9, 10) may potentiate the response to tachykinins released from airway vagal C fibers by histamine (3). These tachykinins can cause bronchoconstriction either directly via actions on smooth muscle or indirectly by increasing release of acetylcholine from parasympathetic nerves (11). We, as well as others (5), have not found smooth muscle hyperresponsiveness to histamine in vitro in tissues from virus-infected animals. Therefore, we favor virus-induced alterations in neural control of airway smooth muscle as the explanation for the findings of Laitinen and colleagues. DAVID B. JACOBY, M.D. Johns Hopkins Asthma & Allergy Center Baltimore, Maryland ALLISON D. FRYER, PH.D.
Johns Hopkins School of Public Health and Hygiene Baltimore, Maryland 1. LaitinenLA, ElkinRB,EmpeyDW, Jacobs L, MillsJ, Nadel JA. Bronchial hyperresponsiveness in normal subjects during attenuated influenza virus infection. Am Rev Respir Dis 1991; 143:358-61. 2. Holtzman MJ, Sheller JR, Dimeo M, Nadel JA, Boushey HA. Effect of ganglionicblockadeon bronchial reactivityin atopic subjects. Am Rev Respir Dis 1980; 122:17-25. 3. Saria A, MartlingC, Yan Z, Theodorsson-Norheim E, Gamse R, Lundberg JM. Release of multipletachykininsfrom capsaicin-sensitive sensory nerves in the lung by bradykinin, histamine, dimethylphenyl piperazinium, and vagal nerve stimulation. Am Rev Respir Dis 1988; 137:1330-5. 4. EmpeyDW, LaitinenLA, Jacobs L, Gold WM,NadelJA. Mechanisms of bronchial hyperreactivity in normal subjectsafter upper respiratorytract infection. Am Rev Respir Dis 1976; 113:131-9. 5. Buckner CK, Songsiridej V, Dick EC, BusseWW. In vivo and in vitro studies on the use of the guinea pig as a model for virus-provoked airway hyperreactivity. Am Rev Respir Dis 1985; 132:305-10. 6. Fryer AD, Jacoby DB.Parainfluenza virus infectiondamagesinhibitory M. muscarinic receptors on pulmonary parasympathetic nerves in the guinea-pig. Br J Pharmacol 1990; 102:267-71. 7. Fryer AD, Maclagan J. Muscarinic inhibitory receptors in pulmonary parasympathetic nerves in the guinea pig. Br J Pharmacol 1984; 83:973-8. 8. FryerAD,El-Fakahany EE, JacobyDB. Parainfluenza virustype1reduces the affinityof agonists for muscarinic receptors in guinea-pig lung and heart. Eur J Pharmacol 1990; 181:51-8. 9. JacobyDB,Tamaoki J, BorsonDB,NadelJA. Influenza infection causes
To the Editor: I read with interest the controversy between Partinen and Hoffstein (1)regarding the relative importance of the body mass index (BMI) and neck circumference in accounting for individual differences in the apnea/hypopnea index (AHl). The real issue underlying use of multiple linear regression is not which variable is entered first or second in the regression but, rather, if the assumptions of the regression are being met. To the extent that many of the key variables entered in the regression were highly correlated with each other (table 2 of the paper [2]), the assumption of absence of colinearity was violated, and any regression performed on these data will be invalid unless such redundancy is first removed. When the 10variables shown in table 2 were reduced by entering the correlation coefficients into a principal components analysis with varimax rotation, four factors emerged accounting for 76.1070 of the cumulative variance. Factor loadings are shown below.
AHI Age BMI EXT LEN PRO MID DIS GLO TRA 070 var
Fac I
Fac II
Fac III
Fac IV
- .15 .07 - .11 -.01 -.10 -.05 .28 .80* .94* .90* (31.2)
.74* .05 .87* .89* .10 -.03 - .19 - .18 - .18 .03 (18.3)
-.08 -.05
.30 .87* - .11 .02 ,42 - .25 .07 .03 - .01 .03 (10.2)
.00 -.06 .57* .74* .87* ,41
.04 - .16 (16,4)
* Primary factor loading of each variable. Note that BMI and external neck circumference (EXT) load on the same factor. This means that within these data there is no way to distinguish the relative importance of BMI and external neck size. The overlap within the data simply will not allow for it. The advent and availability of powerful statistical packages have allowed marvelously rich and complex data sets to be analyzed with relative ease. Unfortunately whether the assumptions underlying use of statistical procedures are met is still very much left to the individual investigator and, as such, must be examined carefully. DONALD
L.
BLIWISE, PH.D.
Stanford University Medical Center Stanford, California 1. Partinen M, Hoffstein V. Letters to the editor. Am Rev Respir Dis 1991;143:204.
2. Katz I, StradlingJ, SlutskyAS, ZamelN, Hoffstein V. Do patients with obstructive sleep apnea have thick necks? Am Rev Respir Dis 1990; 141: 1228-31.
VIRUS-INDUCED AIRWAY HYPERRESPONSIVENESS - POSSffiLE INVOLVEMENT OF NEURAL MECHANISMS
To the Editor: We enjoyed reading Dr. Laitinen's article that demonstrated airway hyperresponsiveness to histamine after infection with live attenuated influenza virus in previously normal subjects (1). Because histamine both stimulates vagally induced reflex bronchoconstriction (2) and releases tachykinins from sensory nerves (3), we suggest that the described virus-induced hyperresponsiveness may be neurally 1422
mediated. Indeed, the same group has shown that hyperresponsiveness to histamine in patients with naturally occurring viral infections can be blocked by atropine (4). Likewise, virus-induced hyperresponsiveness to histamine can be blocked by vagotomy in the guinea pig (5). These observations are consistent with our recent finding that parainfluenza virus infection in guinea pigs selectively damages inhibitory muscarinic receptors on pulmonary parasympathetic nerves (6). These receptors (of the M 2 subtype) normally provide a negative feedback whereby acetylcholine released from the vagus nerve turns off further release of acetylcholine (7). When these receptors do not function (as is the case in virus-infected guinea pigs), the bronchoconstriction caused by electrical stimulation of the vagus is markedly increased and cannot be suppressed by a M 2 agonist such as pilocarpine. This loss of function may be caused by viral neuraminidase (an enzyme present in both influenza and parainfluenza viruses) cleaving off sialic acid residues from the M 2 receptor. We have demonstrated, using ligand binding studies, that agonist affinity for the M 2 receptor is markedly decreased when the receptor is exposed to parainfluenza virus in vitro (8). This effect can be mimicked by an equivalent amount of Clostridium perfringens neuraminidase and blocked by 2,3-dehydro-2-deoxy-N-acetylneuraminic acid, a neuraminidase inhibitor. Alternatively, virus-induced loss of airway neutral endopeptidase activity (9, 10) may potentiate the response to tachykinins released from airway vagal C fibers by histamine (3). These tachykinins can cause bronchoconstriction either directly via actions on smooth muscle or indirectly by increasing release of acetylcholine from parasympathetic nerves (11). We, as well as others (5), have not found smooth muscle hyperresponsiveness to histamine in vitro in tissues from virus-infected animals. Therefore, we favor virus-induced alterations in neural control of airway smooth muscle as the explanation for the findings of Laitinen and colleagues. DAVID B. JACOBY, M.D. Johns Hopkins Asthma & Allergy Center Baltimore, Maryland ALLISON D. FRYER, PH.D.
Johns Hopkins School of Public Health and Hygiene Baltimore, Maryland 1. LaitinenLA, ElkinRB,EmpeyDW, Jacobs L, MillsJ, Nadel JA. Bronchial hyperresponsiveness in normal subjects during attenuated influenza virus infection. Am Rev Respir Dis 1991; 143:358-61. 2. Holtzman MJ, Sheller JR, Dimeo M, Nadel JA, Boushey HA. Effect of ganglionicblockadeon bronchial reactivityin atopic subjects. Am Rev Respir Dis 1980; 122:17-25. 3. Saria A, MartlingC, Yan Z, Theodorsson-Norheim E, Gamse R, Lundberg JM. Release of multipletachykininsfrom capsaicin-sensitive sensory nerves in the lung by bradykinin, histamine, dimethylphenyl piperazinium, and vagal nerve stimulation. Am Rev Respir Dis 1988; 137:1330-5. 4. EmpeyDW, LaitinenLA, Jacobs L, Gold WM,NadelJA. Mechanisms of bronchial hyperreactivity in normal subjectsafter upper respiratorytract infection. Am Rev Respir Dis 1976; 113:131-9. 5. Buckner CK, Songsiridej V, Dick EC, BusseWW. In vivo and in vitro studies on the use of the guinea pig as a model for virus-provoked airway hyperreactivity. Am Rev Respir Dis 1985; 132:305-10. 6. Fryer AD, Jacoby DB.Parainfluenza virus infectiondamagesinhibitory M. muscarinic receptors on pulmonary parasympathetic nerves in the guinea-pig. Br J Pharmacol 1990; 102:267-71. 7. Fryer AD, Maclagan J. Muscarinic inhibitory receptors in pulmonary parasympathetic nerves in the guinea pig. Br J Pharmacol 1984; 83:973-8. 8. FryerAD,El-Fakahany EE, JacobyDB. Parainfluenza virustype1reduces the affinityof agonists for muscarinic receptors in guinea-pig lung and heart. Eur J Pharmacol 1990; 181:51-8. 9. JacobyDB,Tamaoki J, BorsonDB,NadelJA. Influenza infection causes
