- Email: [email protected]
Extracorporeal liver perfusion
RESEARCH LETTERS 2
Mean value (range) or proportion positive
Patient characteristics Duration of HAART, months CD4 number at start of HAART, cells/mm3 Lowest documented CD4, cells/mm3 Highest documented viral load, copies/mL (geometric mean) Duration of suppression to <500 copies/mL, months PCR results Gag RNA PBMC Tissue Multiply-spliced RNA PBMC Tissue
HAART*
HAART+IL-2
Total
(n=13) 20 (11–31)
(n=13) 18 (10–27)
(n=26) 19 (10–31)
307 (10–748)
644 (270–1264)
489 (10–1264)
216 (7–623)
337 (157–552)
277 (7–623) 20 604 (300–307 000)
18 (7–30)
16 (9–25)
17 (7–30)
13/13
11/13 2/2
24/26 2/2
8/13
6/13 1/2
14/26 1/2
Patient characteristics and PCR results
patients, HIV-1 RNA was detected in lymph node or rectal biopsies. While unspliced RNA can represent either genomic or mRNA, multiply spliced RNA is specific for mRNA. With a seminested PCR technique (sensitivity, 50 copies/million cells) multiply spliced RNA was detected in PBMCs from 14 of 26 patients and in one of two tissue samples. Proviral DNA was detected by PCR in the PBMCs of all patients. This study shows that prolonged treatment with HAART does not result in total suppression of viral replication despite suppression of plasma HIV-1 RNA to undetectable levels. HIV-1 RNA was detected in the PBMCs or tissue of all patients. Detection of HIV-1 RNA in PBMCs in the absence of detectable plasma HIV-1 RNA represents intracellular RNA that has been newly synthesised. Detection of multiply-spliced RNA by a less sensitive assay in the majority of patients further supports this conclusion. Although this study was not case-controlled, IL-2 therapy in general did not appear to affect the low levels of HIV-1 replication found in these patients. However, consistent with other reports with sensitive culture methods (Chung T-W, unpublished observations), the two patients with undetectable HIV RNA in PBMCs were both receiving IL-2. Both patients had RNA detectable in lymphoid tissue, supporting the concept that even in these patients there is ongoing, low-level viral replication that may not be detected by culture techniques, although the alternative explanation that the RNA detected by our techniques is from noninfectious virus cannot be excluded. The current stated goal of antiretroviral therapy is totally to inhibit HIV-1 replication and reduce plasma levels to undetectable,1,5 although the term undetectable is imprecise and dependent on the assay being used. Total inhibition may currently be neither possible nor necessary. Since levels of HIV-1 being produced presumably represent a continuum that extends below the detection limits of available assays, making decisions to change antivirals as soon as HIV-1 becomes detectable may not be appropriate. Furthermore, recognition that there is ongoing replication focuses on the need to identify regimens that are more effective in inhibiting HIV-1, and highlights the need for better viral suppression before eradication of HIV-1 can be considered a realistic goal. This study was supported by NCI contract NO1-CO-5600. Centers for Disease Control and Prevention. Report of the NIH Panel to define principles of therapy of HIV and guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. MMWR Morb Mortal Wkly Rep 1998; 47: 1–83.
120
4
5 19 753 21 491 (300–307 000) (<500–307 000)
*Includes 1 patient who discontinued IL-2 one year before beginning HAART.
1
3
Perelson AS, Essunger P, Cao Y, et al. Decay characteristics of HIV-1 infected compartments during combination therapy. Nature 1997; 387: 188–91. Chun TW, Stuyver L, Mizell SB, et al. Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy. Proc Natl Acad Sci USA 1997; 94: 13193–97. Kovacs JA, Vogel S, Albert JM, et al. Controlled trial of interleukin-2 infusions in patients infected with the human immunodeficiency virus. N Engl J Med 1996; 335: 1350–56. Ho DD. Towards HIV eradication or remission: the tasks ahead. Science 1998; 280: 1866–67.
SAIC-Frederick, PO Box B, NCI-FCRDC, Frederick, MD, USA, Laboratory of Immunoregulation, National Institutes of Health, Bethesda, MD; DuPont Circle Physicians Group, PC, Washington, DC; Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD 20892, USA (J A Kovacs; e-mail [email protected])
Extracorporeal liver perfusion L P McChesney, E A Fagan, D L Rowell, J V Del Rio, F Fabrega, M Millis, J W Williams
Acute liver failure of unknown aetiology carries an extremely poor prognosis without liver transplantation. The shortage of donor organs compounds the limited time before complications, such as uncontrollable cerebral oedema, preclude transplantation. Bioartificial liver support systems are being considered for such patients and involve flowing blood or plasma over 50–200 g of hepatocytes.1,2 We investigated the quantity of hepatocytes needed to reverse some metabolic defects seen in acute liver failure of unknown aetiology when the only organ available for a 3year-old boy of 15 kg was a 1200 g liver with 50% fat content from a 75 kg cadaver of a 70-year-old man which was ABO incompatible. This boy developed rapid neurological deterioration to grade IV coma with an intracranial pressure of more than 50 mm Hg, despite osmotic therapy and hyperventilation. As the organ available was considered unusable for implantation, we used the organ in extracorporeal liver perfusion via an oxygenated circuit perfused in a physiological range and in parallel with the child (heparinised circulation every 4 min). We estimated this system provided 40 g/kg of hepatocytes (1200 g350% fat/15 kg) of approximately twice that of the patient’s normal liver (20 g/kg). The boy remained haemodynamically stable throughout 68 h of continuous perfusion. His neurological state and liver function improved. The boy’s liver (at 12 h, 60 h) showed no uptake of isotope by scan (2,6-di-isopropyl iminodiacetic acid) and no increase in liver volume by ultrasound. Within 1 h, his intracranial pressure declined to 12–20 mm Hg with normal cerebral perfusion pressure. By 8 h, pressor requirements decreased and urine output increased to 1–2 mL/kg per h. Connection of lactate concentrations, and metabolic acidosis (around 8 h), and coagulopathy (36 h) required longer perfusion times and progressed slowly over days (table). After 24 h, he became arousable (grade II coma). Despite intravenous Time
Before ECLP -4 h
ECLP
+1 h
ICP (mm Hg) 54 9 Coma (grade) IV IV pH* 7·3 7·35 Anion gap 21 6 Lactate (mmol/L) 4·4 Glucose supplement Yes Yes ProTime INR 14·6 Factor V (%) 4 Factors VII (%) 14
After ECLP
+8 h
+24 h
+36 h +68 h +4 h
+16 h
13 III 7·38 1 2·4 Yes 3·9 8 20
18 II 7·4 4
9 I 7·36 7 2·5 No 2·3 35 32
30 IV 7·29 22 4·4† Yes 2·5
No 2·7 18 24
9 0 7·41 10 1·1 No 1·8 53 36
7 I 7·33 12 No 1·4
ECLP=extracorporeal liver perfusion. INR=international normalised ratio *Corrected for PCO2. †Derived lactate.
THE LANCET • Vol 353 • January 9, 1999
RESEARCH LETTERS
supplements, blood glucose concentrations stabilised only after 24 h of enteral feeding. By 60 h, the child was awake and able to follow verbal commands. Extracorporeal liver perfusion was terminated at 68 h for bleeding complications of an access site. His intracranial pressure remained less than 20 mm Hg during the first 4 h off ECLP. At 8 h, hypoglycaemia and acidosis returned and intracranial pressure rose to 30 mm Hg. Hepatic encephalopathy progressed from grade 0 to IV over the next 16 h. As a suitable cadaveric donor was still not available, family members were screened for compatibility and an orthotopic liver transplant from a living related donor was done. The child awoke within 24 h of transplantation. Histopathology of the explant revealed a collapsed liver with a marked inflammatory response and without evidence or regeneration or clues as to the aetiology of liver failure. We attribute these improvements in this boy to a functioning extracorporeal liver perfusion, since there was no evidence of regeneration of his own liver, and he deteriorated within hours of discontinuing perfusion. Rapid stabilisation of intracranial pressure (within 1 h) and cerebral oedema has also been observed with bioartificial liver support systems.1,2 Our findings are consistent with data extrapolated from animal and human models of hepatic resection that require more than 6 g/kg (30% of normal) of functioning hepatocytes to support metabolic and synthetic features of liver function.3 That the beneficial effect on intracranial pressure extended beyond the perfusion suggests a protective effect that has also been seen with the bioartificial liver support systems (AA Demetriou, personal communication). Comparison of the presented data with current bioartificial liver support systems would suggest that a relatively small hepatocyte mass and limited perfusion times (typically <12 h) may be sufficient to control intracranial pressure, but, a greater mass of hepatocytes and a longer perfusion times may be required to improve metabolic function. We thank V Rizzo, members of the perfusion team, and the nursing staff for their assistance in the care of the patient, and J Cummings for assistance in the preparation of this manuscript. 1 2 3
Rozga J, Podesta L, Demetriou AA. A bioartificial liver to treat severe acute liver failure. Ann Surg 1994; 219: 538–46. Kelly JH, Sussman HL. The hepatic extracoporeal liver assist device in the treatment of fulminant hepatic failure. ASAIO J 1994; 1: 83–85. Pack GT, Islami AH, Hubbard JC, Brasfield RD. Regeneration of human liver after hepatectomy. Surgery 1962; 52: 617–22.
Sections of Transplantation Surgery, and Hepatology (L P McChesney), Rush Medical College, Rush-Presbyterian-St Luke’s Medical Center, Chicago, IL60612, USA; and University of Chicago, Chicago
Edentulism and worsening of obstructive sleep apnoea Caterina Bucca, Stefano Carossa, Sonia Pivetti, Valerio Gai, Giovanni Rolla, Giulio Preti
We reported on an obese man aged 44 years (body-mass index 39 kg/m2) with obstructive sleep apnoea (OSA), who was seen in our respiratory clinic for chronic obstructive lung disease with chronic respiratory failure. The patient presented in September, 1998, for worsening dyspnoea, increasing daytime sleepiness, and a 10 kg gain in body weight. He denied chest pain, increase in cough and sputum production, or haemophtisis. Clinical assessment showed severe cianosis, hypersomnolence, arterial hypertension (160/120 mm Hg), and left and right heart failure. Since his previous visit in April, 1998, he had worsened respiratory failure (decrease in partial pressure of arterial oxygen from 60 mm Hg to 37 mm Hg, increase in partial pressure of
THE LANCET • Vol 353 • January 9, 1999
With dentures Cephalometry Mean (SD) A-Pphw distance (cm) Polysomnography Mean (SD) AHI Mean (SD) Number of obstructive apneas Mean (SD) lowest haemoglobin saturation during sleep % Mean (SD) haemoglobin saturation during sleep %
1·47 (0·11)
Without dentures 1·02 (0·15)
p 0·015
13·0 (4·2) 8·7 (3·1)
20·1 (5·3) 18·0 (4·5)
0·048 0·050
85·0 (2·0)
81·8 (2·4)
0·013
92·2 (2·1)
89·8 (2·0)
0·017
A-Pphw=anteroposterior oropharyngeal wall distance.
Effect of removing dentures on retropharyngeal space and OSA
arterial carbon dioxide from 45 mm Hg to 73 mm Hg and increase in hematocrit from 54% to 60%) and substantial decrease of vital capacity (VC, from 3·7 L to 2·6 L) and forced expiratory volume in 1s (FEV1, from 2·1 L to 1·4 L), with no exacerbation of airway obstruction (FEV1/VC from 57% to 61%). Chest radiography showed enlarged heart and lung congestion. Electrocardiography showed sinusal tachicardia, right-bundle branch block with no sign of recent or past myocardial injury. The only important recent event was that the patient had had all his teeth extracted because of severe periodontal disease. He had been wearing complete dentures for 2 months and had been recommended to remove them during the night. We suspected that the clinical picture depended on the worsening of OSA caused by edentulism. To explore this hypothesis, we did two polysomnographic studies1 on two consecutive nights, one with dentures and the other without dentures. With dentures in, the apnoea plus hypopnoea index (AHI) (number of apnoea/hypopnoea episodes per h) was only slightly higher than that recorded with natural teeth 1 year previously (22·4 vs 16·7), but without dentures the AHI almost doubled (41·1). Supine lateral cephalometry2 showed that the removal of dentures led to a striking decrease in the anteroposterior oropharyngeal wall distance from 1·5 cm to 0·6 cm. These findings were consistent with our hypothesis that edentulism worsened OSA by a decrease in the retropharyngeal space. We extended our study to other patients with OSA and complete dentures. Data from six edentulous men with OSA (mean age 63 [SE 4] years, body-mass index 31 [2] kg/m2) confirm that removal of dentures significantly decreases the retropharyngeal space, and that sleeping without dentures is associated with a significant increase in AHI, and decrease in mean and lowest arterial haemoglobin saturation (table). OSA is a common disorder, especially among elderly people. 61% of patients older than 50 years are estimated to meet the minimum criteria for OSA (AHI >5).1 Although craniomandibular abnormalities are well-recognised risk factors for OSA,1,2 the role of edentulism has never been investigated. This absence of research is surprising, since dental epidemiological and demographic studies3 estimate that edentulism troubles about 18% of patients older than 60 years and that the prevalence will remain constant over the next 30 years. About 10% of elderly people would, therefore, be at risk of worsening OSA as a result of edentulism, as well as related morbidity and mortality.1 Dentists generally recommended removal of dentures during the night, since constant wearing can increase the risk of denture irritations,4 periodontitis, and other risks in overdenture wearers.5 Our findings, if confirmed, suggest that in patients with OSA, the advantages of removing dentures during sleep should be weighted against the risk of worsening upper-airway collapse.
121
Mean value (range) or proportion positive
Patient characteristics Duration of HAART, months CD4 number at start of HAART, cells/mm3 Lowest documented CD4, cells/mm3 Highest documented viral load, copies/mL (geometric mean) Duration of suppression to <500 copies/mL, months PCR results Gag RNA PBMC Tissue Multiply-spliced RNA PBMC Tissue
HAART*
HAART+IL-2
Total
(n=13) 20 (11–31)
(n=13) 18 (10–27)
(n=26) 19 (10–31)
307 (10–748)
644 (270–1264)
489 (10–1264)
216 (7–623)
337 (157–552)
277 (7–623) 20 604 (300–307 000)
18 (7–30)
16 (9–25)
17 (7–30)
13/13
11/13 2/2
24/26 2/2
8/13
6/13 1/2
14/26 1/2
Patient characteristics and PCR results
patients, HIV-1 RNA was detected in lymph node or rectal biopsies. While unspliced RNA can represent either genomic or mRNA, multiply spliced RNA is specific for mRNA. With a seminested PCR technique (sensitivity, 50 copies/million cells) multiply spliced RNA was detected in PBMCs from 14 of 26 patients and in one of two tissue samples. Proviral DNA was detected by PCR in the PBMCs of all patients. This study shows that prolonged treatment with HAART does not result in total suppression of viral replication despite suppression of plasma HIV-1 RNA to undetectable levels. HIV-1 RNA was detected in the PBMCs or tissue of all patients. Detection of HIV-1 RNA in PBMCs in the absence of detectable plasma HIV-1 RNA represents intracellular RNA that has been newly synthesised. Detection of multiply-spliced RNA by a less sensitive assay in the majority of patients further supports this conclusion. Although this study was not case-controlled, IL-2 therapy in general did not appear to affect the low levels of HIV-1 replication found in these patients. However, consistent with other reports with sensitive culture methods (Chung T-W, unpublished observations), the two patients with undetectable HIV RNA in PBMCs were both receiving IL-2. Both patients had RNA detectable in lymphoid tissue, supporting the concept that even in these patients there is ongoing, low-level viral replication that may not be detected by culture techniques, although the alternative explanation that the RNA detected by our techniques is from noninfectious virus cannot be excluded. The current stated goal of antiretroviral therapy is totally to inhibit HIV-1 replication and reduce plasma levels to undetectable,1,5 although the term undetectable is imprecise and dependent on the assay being used. Total inhibition may currently be neither possible nor necessary. Since levels of HIV-1 being produced presumably represent a continuum that extends below the detection limits of available assays, making decisions to change antivirals as soon as HIV-1 becomes detectable may not be appropriate. Furthermore, recognition that there is ongoing replication focuses on the need to identify regimens that are more effective in inhibiting HIV-1, and highlights the need for better viral suppression before eradication of HIV-1 can be considered a realistic goal. This study was supported by NCI contract NO1-CO-5600. Centers for Disease Control and Prevention. Report of the NIH Panel to define principles of therapy of HIV and guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. MMWR Morb Mortal Wkly Rep 1998; 47: 1–83.
120
4
5 19 753 21 491 (300–307 000) (<500–307 000)
*Includes 1 patient who discontinued IL-2 one year before beginning HAART.
1
3
Perelson AS, Essunger P, Cao Y, et al. Decay characteristics of HIV-1 infected compartments during combination therapy. Nature 1997; 387: 188–91. Chun TW, Stuyver L, Mizell SB, et al. Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy. Proc Natl Acad Sci USA 1997; 94: 13193–97. Kovacs JA, Vogel S, Albert JM, et al. Controlled trial of interleukin-2 infusions in patients infected with the human immunodeficiency virus. N Engl J Med 1996; 335: 1350–56. Ho DD. Towards HIV eradication or remission: the tasks ahead. Science 1998; 280: 1866–67.
SAIC-Frederick, PO Box B, NCI-FCRDC, Frederick, MD, USA, Laboratory of Immunoregulation, National Institutes of Health, Bethesda, MD; DuPont Circle Physicians Group, PC, Washington, DC; Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD 20892, USA (J A Kovacs; e-mail [email protected])
Extracorporeal liver perfusion L P McChesney, E A Fagan, D L Rowell, J V Del Rio, F Fabrega, M Millis, J W Williams
Acute liver failure of unknown aetiology carries an extremely poor prognosis without liver transplantation. The shortage of donor organs compounds the limited time before complications, such as uncontrollable cerebral oedema, preclude transplantation. Bioartificial liver support systems are being considered for such patients and involve flowing blood or plasma over 50–200 g of hepatocytes.1,2 We investigated the quantity of hepatocytes needed to reverse some metabolic defects seen in acute liver failure of unknown aetiology when the only organ available for a 3year-old boy of 15 kg was a 1200 g liver with 50% fat content from a 75 kg cadaver of a 70-year-old man which was ABO incompatible. This boy developed rapid neurological deterioration to grade IV coma with an intracranial pressure of more than 50 mm Hg, despite osmotic therapy and hyperventilation. As the organ available was considered unusable for implantation, we used the organ in extracorporeal liver perfusion via an oxygenated circuit perfused in a physiological range and in parallel with the child (heparinised circulation every 4 min). We estimated this system provided 40 g/kg of hepatocytes (1200 g350% fat/15 kg) of approximately twice that of the patient’s normal liver (20 g/kg). The boy remained haemodynamically stable throughout 68 h of continuous perfusion. His neurological state and liver function improved. The boy’s liver (at 12 h, 60 h) showed no uptake of isotope by scan (2,6-di-isopropyl iminodiacetic acid) and no increase in liver volume by ultrasound. Within 1 h, his intracranial pressure declined to 12–20 mm Hg with normal cerebral perfusion pressure. By 8 h, pressor requirements decreased and urine output increased to 1–2 mL/kg per h. Connection of lactate concentrations, and metabolic acidosis (around 8 h), and coagulopathy (36 h) required longer perfusion times and progressed slowly over days (table). After 24 h, he became arousable (grade II coma). Despite intravenous Time
Before ECLP -4 h
ECLP
+1 h
ICP (mm Hg) 54 9 Coma (grade) IV IV pH* 7·3 7·35 Anion gap 21 6 Lactate (mmol/L) 4·4 Glucose supplement Yes Yes ProTime INR 14·6 Factor V (%) 4 Factors VII (%) 14
After ECLP
+8 h
+24 h
+36 h +68 h +4 h
+16 h
13 III 7·38 1 2·4 Yes 3·9 8 20
18 II 7·4 4
9 I 7·36 7 2·5 No 2·3 35 32
30 IV 7·29 22 4·4† Yes 2·5
No 2·7 18 24
9 0 7·41 10 1·1 No 1·8 53 36
7 I 7·33 12 No 1·4
ECLP=extracorporeal liver perfusion. INR=international normalised ratio *Corrected for PCO2. †Derived lactate.
THE LANCET • Vol 353 • January 9, 1999
RESEARCH LETTERS
supplements, blood glucose concentrations stabilised only after 24 h of enteral feeding. By 60 h, the child was awake and able to follow verbal commands. Extracorporeal liver perfusion was terminated at 68 h for bleeding complications of an access site. His intracranial pressure remained less than 20 mm Hg during the first 4 h off ECLP. At 8 h, hypoglycaemia and acidosis returned and intracranial pressure rose to 30 mm Hg. Hepatic encephalopathy progressed from grade 0 to IV over the next 16 h. As a suitable cadaveric donor was still not available, family members were screened for compatibility and an orthotopic liver transplant from a living related donor was done. The child awoke within 24 h of transplantation. Histopathology of the explant revealed a collapsed liver with a marked inflammatory response and without evidence or regeneration or clues as to the aetiology of liver failure. We attribute these improvements in this boy to a functioning extracorporeal liver perfusion, since there was no evidence of regeneration of his own liver, and he deteriorated within hours of discontinuing perfusion. Rapid stabilisation of intracranial pressure (within 1 h) and cerebral oedema has also been observed with bioartificial liver support systems.1,2 Our findings are consistent with data extrapolated from animal and human models of hepatic resection that require more than 6 g/kg (30% of normal) of functioning hepatocytes to support metabolic and synthetic features of liver function.3 That the beneficial effect on intracranial pressure extended beyond the perfusion suggests a protective effect that has also been seen with the bioartificial liver support systems (AA Demetriou, personal communication). Comparison of the presented data with current bioartificial liver support systems would suggest that a relatively small hepatocyte mass and limited perfusion times (typically <12 h) may be sufficient to control intracranial pressure, but, a greater mass of hepatocytes and a longer perfusion times may be required to improve metabolic function. We thank V Rizzo, members of the perfusion team, and the nursing staff for their assistance in the care of the patient, and J Cummings for assistance in the preparation of this manuscript. 1 2 3
Rozga J, Podesta L, Demetriou AA. A bioartificial liver to treat severe acute liver failure. Ann Surg 1994; 219: 538–46. Kelly JH, Sussman HL. The hepatic extracoporeal liver assist device in the treatment of fulminant hepatic failure. ASAIO J 1994; 1: 83–85. Pack GT, Islami AH, Hubbard JC, Brasfield RD. Regeneration of human liver after hepatectomy. Surgery 1962; 52: 617–22.
Sections of Transplantation Surgery, and Hepatology (L P McChesney), Rush Medical College, Rush-Presbyterian-St Luke’s Medical Center, Chicago, IL60612, USA; and University of Chicago, Chicago
Edentulism and worsening of obstructive sleep apnoea Caterina Bucca, Stefano Carossa, Sonia Pivetti, Valerio Gai, Giovanni Rolla, Giulio Preti
We reported on an obese man aged 44 years (body-mass index 39 kg/m2) with obstructive sleep apnoea (OSA), who was seen in our respiratory clinic for chronic obstructive lung disease with chronic respiratory failure. The patient presented in September, 1998, for worsening dyspnoea, increasing daytime sleepiness, and a 10 kg gain in body weight. He denied chest pain, increase in cough and sputum production, or haemophtisis. Clinical assessment showed severe cianosis, hypersomnolence, arterial hypertension (160/120 mm Hg), and left and right heart failure. Since his previous visit in April, 1998, he had worsened respiratory failure (decrease in partial pressure of arterial oxygen from 60 mm Hg to 37 mm Hg, increase in partial pressure of
THE LANCET • Vol 353 • January 9, 1999
With dentures Cephalometry Mean (SD) A-Pphw distance (cm) Polysomnography Mean (SD) AHI Mean (SD) Number of obstructive apneas Mean (SD) lowest haemoglobin saturation during sleep % Mean (SD) haemoglobin saturation during sleep %
1·47 (0·11)
Without dentures 1·02 (0·15)
p 0·015
13·0 (4·2) 8·7 (3·1)
20·1 (5·3) 18·0 (4·5)
0·048 0·050
85·0 (2·0)
81·8 (2·4)
0·013
92·2 (2·1)
89·8 (2·0)
0·017
A-Pphw=anteroposterior oropharyngeal wall distance.
Effect of removing dentures on retropharyngeal space and OSA
arterial carbon dioxide from 45 mm Hg to 73 mm Hg and increase in hematocrit from 54% to 60%) and substantial decrease of vital capacity (VC, from 3·7 L to 2·6 L) and forced expiratory volume in 1s (FEV1, from 2·1 L to 1·4 L), with no exacerbation of airway obstruction (FEV1/VC from 57% to 61%). Chest radiography showed enlarged heart and lung congestion. Electrocardiography showed sinusal tachicardia, right-bundle branch block with no sign of recent or past myocardial injury. The only important recent event was that the patient had had all his teeth extracted because of severe periodontal disease. He had been wearing complete dentures for 2 months and had been recommended to remove them during the night. We suspected that the clinical picture depended on the worsening of OSA caused by edentulism. To explore this hypothesis, we did two polysomnographic studies1 on two consecutive nights, one with dentures and the other without dentures. With dentures in, the apnoea plus hypopnoea index (AHI) (number of apnoea/hypopnoea episodes per h) was only slightly higher than that recorded with natural teeth 1 year previously (22·4 vs 16·7), but without dentures the AHI almost doubled (41·1). Supine lateral cephalometry2 showed that the removal of dentures led to a striking decrease in the anteroposterior oropharyngeal wall distance from 1·5 cm to 0·6 cm. These findings were consistent with our hypothesis that edentulism worsened OSA by a decrease in the retropharyngeal space. We extended our study to other patients with OSA and complete dentures. Data from six edentulous men with OSA (mean age 63 [SE 4] years, body-mass index 31 [2] kg/m2) confirm that removal of dentures significantly decreases the retropharyngeal space, and that sleeping without dentures is associated with a significant increase in AHI, and decrease in mean and lowest arterial haemoglobin saturation (table). OSA is a common disorder, especially among elderly people. 61% of patients older than 50 years are estimated to meet the minimum criteria for OSA (AHI >5).1 Although craniomandibular abnormalities are well-recognised risk factors for OSA,1,2 the role of edentulism has never been investigated. This absence of research is surprising, since dental epidemiological and demographic studies3 estimate that edentulism troubles about 18% of patients older than 60 years and that the prevalence will remain constant over the next 30 years. About 10% of elderly people would, therefore, be at risk of worsening OSA as a result of edentulism, as well as related morbidity and mortality.1 Dentists generally recommended removal of dentures during the night, since constant wearing can increase the risk of denture irritations,4 periodontitis, and other risks in overdenture wearers.5 Our findings, if confirmed, suggest that in patients with OSA, the advantages of removing dentures during sleep should be weighted against the risk of worsening upper-airway collapse.
121