occluded his right main bronchus revealing patent distal bronchi. The patient was subsequently extubated and completed an uneventful postoperative course. DISCUSSION

Since the description of subclavian cannulation by Aubaniac,3 central line catheters have become indispensable in medical care. Catheter-related risks are legion, but serious sequelae fortunately are rare. Most of the structures surrounding the central vein have been involved with complications. 4•5 However, a cava-bronchial fistula has not been reported in a computerized search of the Medline database. Central vein perforation has occurred with polyvinyl catheters, although modifications of their tips may reduce the incidence. Silicone rubber catheters with their pliability and softness have even a lower incidence ofcatheter erosion. 6 In this case, several features may have worked in tandem to produce this potentially lethal complication. First, the catheter tip, although placed appropriately in the superior vena cava, was cut at a taper. Studies have shown, despite fixation of the catheter, that the tip may still move, as much as several centimeters. 7 Second, with a tapered tip and infusion of an irritating substance (fluorouracil), continued contact with the central vein wall may have led to erosion. In management of this unusual fistula, the diagnosis was confirmed by intracatheter contrast injection. Bronchoscopy was helpful once the airway was stabilized to confirm the probable location and also to place an endobronchial blocker. With thoracotom~ control of the vessel can be assured and the fistula divided. Healing in general should not be a difficult problem as there is only inflammatory tissue and no epithelialized fistulous tract. The placement of additional tissue to separate the vessels may help reduce any potential recurrence. Central lines will continue to produce complications. 8 Most present shortly after the procedure and can be treated appropriately with minimal delay in diagnosis. Rarely will complications present months later. In this situation, their temporal relationship with the placement of the catheter is obscured and the diagnosis may be more difficult. N evertheless, when patients with indwelling central catheters develop hemoptysis, catheter-related bronchial perforation should be considered as a possible cause. REFERENCES

1 Wagman LD, Kirkemo A, Johnston MR. Venous access: a prospective, randomized study of the Hickman catheter. Surgery 1984; 95:303-08 2 Ellis LM, Vogel SB, Copeland EM III. Central venous catheter erosions. Ann Surg 1989; 209:475-78 3 Aubaniac R. L'injection intraveineuse sous-elaviere. Presse Med 1952; 60:1656 4 Clarke DE, Raffin TA. Infectious complications of long-term central venous catheters. Chest 1990; 97:966-72 5 Black A, Uoyd-Thomas AR. An unusual case of stridor. Anaesthesia 1988; 43:870-71 6 Cathcart-Rake WF, Mowery WE. Intrapericardial infusion of 58uorouracil. Cancer 1991; 67:735-37 7 Krasnow SH, Rhodes G, Boyer M, Criton ML, et a1. Hickman catheter tip displacement. South Med J 1985; 78: 1327-29 8 Kaye CG, Smith DR. Complications of central venous catheterization. 8M} 1988; 297:572-73

1288

Ventilatory Dysfunction in Severe Anorexia Nervosa* C. Francis Ryan, M. B., F. C. C. ~;t J Scott Whittaker, M. D.; and

Jeremy D. Road, M.D.

A 25-year-old woman suffering from chronic anorexia nervosa lost more than 50 percent of her body weight and presented with generalized muscle weakness. Pulmonary function tests showed a severe restrictive defect, and she had marked impairment of respiratory muscle strength and endurance, peripheral muscle function, and hypercapnic ventilatory responses, all of which improved following refeeding. The interaction and response to treatment of these effects on respiratory function are discussed. (Chest 1992; 102:1286-88) FIOIF50 = ratio of force generation following ulnar stimulation at 10 Hz over 50 HZ; 4G ATP =free-energy change of ATP hydrolysis; PO. I = inspiratory mouth pressure lOOms after commencement of randomly occluded breath; PETCO. = endtidal carbon dioxide pressure; Pm = mean mouth pressure ~ norexia

nervosa can cause severe malnutrition resultin~ in significant morbidity and mortality. I Malnutrition has important adverse effects on the respiratory system, 2 particularly on respiratory muscle function,3-5 that may predispose to pulmonary complications. We describe the effects of severe malnutrition on respiratory and peripheral muscle function and control of breathing in a patient with anorexia nervosa.

1'1

CASE REPORT

A 25-year-old woman was admitted for mana~ement of severe malnutrition resulting from chronic anorexia nervosa of 8 years' duration. She complained of prof(lund fatigue and inability to maintain a seated position because of weakness. She had restricted her food intake severely but denied diuretic, laxative, or ipecac abuse. 6 She had had secondary amenorrhea for six years, and bone densitometry one year previously had shown osteopenia. There was no history of chest infection. On examination the patient was markedly cachectic, with reduced secondary sexual characteristics and hilateral parotid swellin~. She weighed 25 kg (46 percent of ideal body weighF), and her height was 163 cm. The heart rate was 48 beats per minute, supine blood pressure was 80140 mm Hg, and the respiratory rate was 100min, with regular and shallow respiration. She had bilateral pitting ankle edema. The breath sounds were normal. The patient had marked proximal muscle weakness and was unable to raise her head from the bed when supine or perform more than three squats from a standing position. The results of laboratory investigations were as follows: hemoglobin, 8.9 Wdl; leukocyte count, 4,700/cu mm; sodium, 135 mEq/L; chloride, 98 mEq/L; potassium, 4.5 mEq/L; bicarbonate, 26 mEq/L; urea, 33 mwdl; creatinine, 0.4 mg/dl; calcium, 8.7 mgldl; phosphate, 1.8 mwdl; magnesium, 1.88 mgldl; total protein, 4.9 gldl; and albumin, 3.0 gldl. An electrocardiogram showed sinus bradycardia, and a chest roentgenogram was normal. After the patient had given informed consent, the follOwing investigations were performed before and after a period of refeeding. *From the Department of Medicine, University of British Columbia, Vancouver, Canada. tBritish Columbia Lung Association Research Fellow. Supported by the British Columbia Lung Association. Reprint requests: Dr. Road, University Hospital, 2211 Wesbrvok Mall, Vancouver, Be, Canada V6T 2B5 Ventilatory Dysfunction in Severe Anorexia Nervosa (Ryan, Whittaker, Road)

Table I-Pulmonary, Respiratory Muscle, and Adductor PoUicis Function Before Refeeding

Data FVC, L FEV" L FEV/FVC IC, L ER~ L FRC,L R~ L TLC,L RVffLC DNA* MI~ cm 1120 ME~ cm H 2O Pm, cm H 20t FIo/FSO, MRR, percent force loss/IO ms§

%+

Value 1.34

1.30 0.97 1.12 0.40 2.70 2.30 3.82 0.60 4.0 34

35 4.0

so

Percent of Predicted

35 39 48 28 98 171 75 80 37 23 13

7.2

After Refeeding

Value 2.31 2.19 0.95 1.48 0.83 2.83 2.00 4.31 0.47 4.0 52 40

8.0 36 9.3

Percent of Predicted 60 65 64 58

102 150

85

80 57 26 26

*DNA, Carhon monoxide diffusing capacity as ratio of alveolar volume, corrected for hemoglobin concentration. tPm, Mean mouth pressure during final increment of threshold loading respiratory muscle endurance test. tFlo/FSO, ratio of force generation following ulnar nerve stimulation at 10 Hz over 50 Hz (normal range, 29±6 percent). §~fRR, Maximal relaxation rate following ulnar nerve stimulation at 30 liz (normal range, 9.6 ± 0.9 percent). Pulmonary function tests (Cybermedic Pulmonary Function System; Spinnaker, 1987; Software 3.04) showed a severe restrictive defect with an increased RV (Table 1). Inspiratory capacity was 48 percent of predicted, and RV was 171 percent of predicted. Diffusing capacity (D) was normal (80 percent of predicted) when corrected for alveolar volume and hemoglobin concentration. Respiratory muscle strength measured as MIP and MEfM was markedly reduced, even when corrected for lung volume9 (Table 1). Inspiratory muscle endurance measured as mean Inouth pressure (Pm) during the final increlnent of a threshold loading test lll was also reduced compared with puhlished normal values." Adductor pollicis muscle testing by supramaximal stimulation of the ulnar nerve l2 showed an increase in relative fclfce generation at low-frequency stimulation (FIOHzI FSOHz =.50 percent) and slowing of the maximal relaxation rate fc)llo\\'ing stimulation at 30 Hz (Table 1). Peak end-tidal CO 2 pressure (PETC0 2) prior to rehreathing was low (Table 2). Hypercapnic ventilatory responses were markedly impaired, both in tenns of minute ventilation'3 and mouth occlusion pressure (Po. I)" (Table 2). Resting energy expenditure (MMC Horizon~ Sensonnedics) was 10\\', and carbon dioxide production (Vcoz) was only 81 mVmin (t'lble 2). A nasoenteric feeding tube was inserted under fluoroscopic visualization, and the patient was refed. She gained 5 kg over four weeks, her weight increasing to 55 percent of ideal body weight. lIer strength increased, she resumed walking unaided, and the peripheral edema resolved. Repeat measurements showed nonnal blood electrolyte level and adductor pollicis muscle function, increased PETC0 2 prior to rehreathing, and large improvements in flo\\' rates, lung volumes, respiratory muscle strength and endurance, and hypercapnic ventilatory responses (Tables 1 and 2). Carbon dioxide production and resting energy expenditure also increased (Tahle 2). DISCUSSION

The adverse effects of malnutrition on respiratory muscle

function have been reported previousl~5 In this patient the severity of malnutrition was notable, and accordingly, the respiratory muscle impairment was more marked. Indeed, previous studies have not shown reductions in TLC below 80 percent of predicted}S In this patient the TLC was only 75 percent of predicted, presumably due to the inability of the weak inspiratory muscles to overcome the normal elastic recoil of the respiratory system. This patient also had markedly impaired ventilatory responses to CO2 , which improved following refeeding. Previously, hypoxic ventilatory responses have been shown to be decreased during semistarvation; 16 however, hypercapnic responses were not similarly affected. The impaired hypercapnic ventilatory responses in our patient may have been due entirely to respiratory muscle weakness; however, the impairment of occlusion pressure responses {PO. I) suggests an additional decrease in respiratory motoneuron activity. The PO. I response to hypercapnia tends to be preserved, even when there is marked weakness of the inspiratory muscles. 17 The ventilatory response to carbon dioxide is preserved in subjects during partial paralysis, with its attendant moderately severe respiratory muscle weakness. 18 Severe malnutrition, therefore, may have impaired the hypercapnic ventilatory response in our patient by decreasing respiratory motoneuron output; however, caution is warranted in interpreting the PO.I responses, as the improvement following weight gain was small compared with the minute ventilation response. It is possible that at very low levels of respiratory drive, the improvements in hypercapnic ventilatory responses in terms of minute ventilation and PO.l are not linearly related. Respiratory failure resulting in hypercapnia does not usually occur until respiratory muscle strength is less than 30 percent of predicted values. In anorexia nervosa and other causes of starvation, it is possible that the reduced energy expenditure, Vco2 , and oxygen consumption {VoJ might delay the onset of hypercapnia. We did not measure arterial blood gas levels in this patient, but the abnormally

Table 2-Hypercapnic Ventilatory Responses and Energy Expenditure Data PETC02ambinlt , mm Hg*

Nonnal (Mean±SD)

~VE!~PETC02'

40±5 2.6± 1.2

~PO.l/~PETC02'

0.5±0.2

Before Refeeding

After Refeeding

0.13

32 0.38

0.06

0.09

26

Uminlmm Hg

cm H 20/mm Hg At PETC0 2 = 55 mm Hg VT,ml F, breaths per min VE,L PO.I, cm H 2O Vc0 2 , mVmin \;0 2, mVmin RQ REE, kcalt

360

26.9±9.2 13.2±0.8 155+ 190+ 0.82+ 1,325+

15 5.4 3.0 81 85

0.95 610

900

16 14.4 3.0 99 104 0.95 760

*Prior to rebreathing. tREE, Resting energy expenditure. tDerived from Harris-Benedict and Weir equations, assuming nonnal weight. CHEST I 102 I 4 I OCTOBER, 1992

1287

low PETC02 prior to rebreathing, in the absence of hyperventilation, indicates that she was hypocapnic. Refeeding causes energy expenditure, Vco 2 , and V02 to increase, leading to increased ventilatory demands, but in our patient the increase in ventilatory capacity after refeeding far exceeded the increase in Vc02. This suggests that, in patients with anorexia nervosa, unless some cause of increased VCO2 such as sepsis, is present, refeeding can be undertaken without the risk of precipitating ventilatory failure. This patient had the abnormalities of adductor pollicis muscle function which have been described previously in severe anorexia nervosa,12 namely: a leftward shift of the force-frequency curve and a slowing of the relaxation rate. These abnormalities are believed to relate to a fall in muscle creatine phosphate levels and free-energy change of ATP hydrolysis (4G ATP)' 19 A fall in 4G ATP has also been closely related to slowing of the relaxation rate in fatigued muscle. 20 Like Russell and associates,12 we demonstrated recovery of adductor pollicis function after four weeks of refeeding; however, although adductor pollicis function improved, abnormalities of respiratory muscle strength and endurance remained. We have previously found normal adductor pollicis function in moderately malnourished patients with COPD who had reduced respiratory muscle strength and endurance. 21 These findings indicate that the functional abnormalities of the adductor pollicis muscle tested by ulnar nerve stimulation occur when malnutrition is severe and re