Alveolar-Arterial Oxygen Gradients Versus the Neonatal Pulmonary Insuff iciency Index for Prediction of Mortal i ty in

ECMO Candidates

By Thomas M. Krummel, Lazar J. Greenfield, Barry V. Kirkpatrick, Dawn G. Mueller, Kathryn W. Kerkering, Miguel Ormazabal, Anthony Napolitano, and Arnold M. Salzberg

Richmond, Virginia

�9 Current selection criteria necessary for intel- ligent application of extracorporeal membrane oxygenation (ECMO) in hypoxic neonates remains controversial. Both the Neonatal Pulmonary Insuffi- ciency Index (NPII) and serial alveolar-arterial oxy- genation gradient measurements (A-a)Do 2 have been recommended. Accordingly, an analysis of 50 con- secutive severely hypoxic neonates was undertaken to assess the predictive value of (A-a)Do z determina- tions and NPII in discriminating survivors from non- survivors. These infants with meconium aspiration syndrome (MAS), congenital diaphragmatic hernia (CDH), or persistent pulmonary hypertension of the newborn (PPHN) required maximum mechanical ven- tilation for hypoxia. Pharmacologic manipulation of pulmonary vascular resistance was attempted in 83%. If postductal (A-aIDo~ remained >--620 torr despite 12 hours of maximum medical therapy, mor- tality was 100%; however, 35% of nonsurvivors were unfortunately excluded. (A-a)Do 2 _>_ 600 tort for 12 hours demonstrated 93.8% mortality, and only 12% of all mortalities were thus excluded. Among nonsurvivors successfully hyperventilated, the NPII could not predict mortality. Ideal selection criteria must exclude those who would otherwise survive without ECMO, yet allow early accurate identifica- tion of the neonate certain to die. It would appear that serial (A-a)Do= determinations best permit this identification and thus orderly application of ECMO.

INDEX WORDS: Alveolar-arterial oxygen gradients; neonatal pulmonary insufficiency index; ECMO.

S INCE Bartlett's report in 1977,1 extracor- poreal membrane oxygenation (ECMO)

has been available to assist certain critically hypoxic newborns dying despite optimal conven- tional management. Application of ECMO can

From the Department of Surgery, Divisions of Pediatric and Cardiothoracic Surgery, Department of Pediatrics, Division of Neonatology Medical College of Virginia, Rich- mond, VA.

Presented at the 32nd Annual Meeting of the Surgical Section of the American Academy of Pediatrics, San Fran- cisco, California, October 22-23, 1983.

Address reprint requests to Arnold M. Salzberg, MD, Box 15, MCV Station, Richmond, VA 23298.

�9 1984 by Grune & Stratton, lnc. 002~3468/84/1904-0010503.00/0

provide life-saving support with quality survival in selected moribund infants. 2-5 The ideal criteria for mortality prediction in hypoxic newborns remain elusive, however, and current methods for the rational selection of ECMO candidates continue to be controversial. Both the Neonatal Pulmonary Insufficiency Index (NPII) 6'7 and serial alveolar-arterial oxygen gradients (A-a)D028-1~ have been used to predict mortality and permit the employment of membrane oxy- genation before mortality becomes a reality.

In order to compare the predictive value of the NPII and serial (A-a)D02, a review of 50 consec- utive near-term infants with severe hypoxia was undertaken to assess the merits of each in dis- criminating survivors from nonsurvivors. Prema- ture infants were not considered ECMO candi- dates and were therefore not included in this review.

MATERIALS AND METHODS

Fifty near-term newborns treated for severe hypoxia at the Medical College of Virginia between May 1977 and Febru- ary 1983 were reviewed retrospectively. The diagnoses included meconium aspiration syndrome (MAS), transient tachypnea of the newborn (TTNB), neonatal asphyxia, pneu- monia, congenital diaphragmatic hernia (CDH), and pri- mary persistent pulmonary hypertension of the newborn (PPHN) (Table 1); infants with cyanotic congenital heart lesions were carefully excluded on the basis of echocardiogra- phy. All neonates were ventilated with an inspired oxygen concentration of 100% (Fio 2 = 1.0), to maintain a postductal PO2 >- 80 torr. Other ventilatory parameters were manipu- lated to optimize arterial blood gases: peak inspiratory pres- sures (PIP) 20 to 60 cm H20; positive end-expiratory pres- sure (PEEP) 0 to 2 cm H20; ventilatory rates 30 to 150 breaths/minute. A number of infants were sedated with morphine or paralyzed with pancuronium.

The clinical diagnosis of pulmonary artery hypertension was made when any of the following criteria were observed in conjunction with a compatible history of physical examina- tion: echocardiographic visualization of a patent foramen ovale (PFO) or demonstration of an atrial shunt with echo- cardiographic contrast; or evidence of a transductal shunt of >__10 mmHg partial pressure of oxygen. This diagnosis was usually confirmed by a positive hyperoxic hyperventilation test.

If PPHN was documented, the infants were hyperventi-

380 Journal of Pediatric Surgery, Vol. 19, No. 4 (August), 1984

MORTALITY PREDICTION IN ECMO CANDIDATES 381

Table 1. MCV Experience With Hypoxic Term Neonates

1977-1983

Meconium Aspiration Syndrome (MAS) 22 Transient Taehypnea of the Newborn ('VI'NB) 3 Congenital Diaphragmatic Hernia (CDH) 9 Neonatal Asphyxia 2 Pneumonia 3 Primary Persistent Pulmonary Hypertension of the

Newborn (PPHN) 11 Total 50

lated to maintain a PCO 2 ~ 30 m m H g and an arterial pH _> 7.50. Systemic arterial blood pressure was augmented with volume expansion and if necessary, dopamine (5 to 15 mcg /kg /min ) . If hyperventilation did not improve oxygena- tion, tolazoline (1 mg / kg bolus, followed by 1 to 4 m g / k g / h r infusion) was administered (83%). Serial postductal determi- nations of (A-a)Do2* were made and compared to the NPII data tabulated. NPII calculations require a simultaneous graph of arterial pH and Fio2 (Fig 1). Normally the plot of Fio2 will rest below that of the pH. If these lines intersect, serious respiratory distress is present. The NPII score repre- sents the integrated area between the two graphs when the Fio2 plot lies above the pH plot.

RESULTS

A total of 50 newborns were reviewed (Table 2). There was no statistically significant difference between survivors and nonsurvivors in weight, sex, gestational age, or A P G A R scores (Table 3). Thirty-eight patients were managed con- ventionally. There was evidence of pulmonary hypertension complicating the underlying disease or existing as the pri- mary pathology in all. Of these, 22 were seen prior to E C M O and 16 after ECM O capability. In this last group, 15 did not meet ECMO criteria and the one infant who did was excluded because of simultaneous perfusion in another patient. In these 16 patients, 6 survived. Twelve additional patients were placed on E C M O after maximal medical therapy was exhausted (Table 2).

When comparing survivors with nonsurvivors using NPII calculations, there is clearly no difference in NPII scores at 12 and 24 hours. Based on these data, NPII calculations are not predictive of mortality. Fig 2 demonstrates a typical NPII graph and illustrates its limitation when successful hyperven- tilation is achieved. The resultant alkalosis precludes signifi- cant accumulation of NPII points. Survivors and nonsurvi- vors have similar plots and no discrimination can be made between them.

(A-a)Do2 computations were more effective in mortality prediction. If the postductal (A-a)DO 2 remained in excess of 620 torr for 12 hours, mortality was 100%. However, this criteria excluded 35% of nonsurvivors from additional thera- py. If the (A-a)Do 2 exceeded 600 tort for 12 hours, mortality was 93.8% and only 12% were excluded.

There were 16 survivors treated without ECMO. No survivor had an (A-a)Do2 >_ 620 torr for a period of 12 hours

*(A-a)Do2 = 760 - (Pao~ + Paco 2 + 47)

. p , , . , . , , . e l h ~ . ~ , , . ~ . . , ~ e ( . o = . J

Fig 1. NPII (A} Normal infant: simultaneous graph of pH and Fio 2. (B) Infant with incipient respiratory distress. (C) Serious respiratory distress occurs when the pH graph crosses the Fio z graph. (D) NPII score = 16 after 4 hours, (E) NPI! score ~ 36 after 20 hours.

or more during the hospital course. One survivor (6.2%) had an (A-a)Do 2 >_ 600 torr for a single 12-hour period.

Accordingly, an (A-a)Do2 >_ 600 torr for over 12 hours despite maximum medical therapy appears capable of pre- dicting a 94% mortality and, equally significant, rejects a very low percentage of the nonsurvivors. The statistical efficacy of serial (A-a)Do2 >_ 600 torr is depicted in Table 4. Mortality is predicted with a sensitivity of 0.888 and with a specificity of 0.933. The accuracy of positive prediction is 0.960 and the accuracy of negative prediction is 0.823. '4

DISCUSSION

The hypoxic neonate has always been a pecu- liar challenge to physicians because of the pro- found changes in cardiopulmonary circulation which surround birth. The normal fetal circula- tory pattern is depicted in Fig 3. High pulmonary vascular resistance in the unexpanded fetal lung leads to equalization of pressures in the pulmo- nary artery and aorta. Most venous blood is diverted at atrial or ductal levels, thus pulmonary

Table 2. MCV Experience With Hypoxic Term Neonates

1977-1983

Standard Treatment + ECMO

(survivors/total) Meconium Aspiration

Syndrome (MAS) 5/15 4/7 Transient Tachypnea

of the Newborn ('I-I'NB) 1/3 Congenital Diaphragmatic

Hernia (CDH) 1/6 2/3 Neonatal Asphyxia I /2 Pneumonia 1/3 Primary Persistent

Pulmonary Hypertension of the Newborn (PPHN) 5/9 1/2 Total 14/38 7/12

382

Table 3. Characteristics of Survivors and Nonsurvivors

Survivors Nonsurvivors N = 21 N = 29

Weight 2 .68 _+ .79 2 .96 _+ .89 NS

Sex 12M:13F 10M:6F NS*

Gestational Age 39.1 _+ 3.0 38 .6 _+ 2.7 NS

APGAR 1 min 5.2 _+ 2.1 3.5 -+ 2.0 NS*

APGAR 5 rain 7.2 +_ 1.5 5.2 _+ 2.6 NS*

* 0 . 1 > P > . 0 5

blood flow is only 10% of total cardiac output. At birth, placental gas exchange is, of course, lost. However, as the lung expands, pulmonary vascu- lar resistance falls, allowing increased perfusion of the lung, now the site of gas exchange. Physio- logic and ultimately anatomic closure of ductal and atrial shunts follows, placing the pulmonary and systemic circulations in series in the normal newborn (Fig 4).

These complex physiologic changes necessary for extrauterine life become inextricably em- meshed with and are often perturbed by any concurrent pathologic process arising in a new- born. Contemporary understanding of such an infant began with Gersony's 1~ recognition of a syndrome of characterized by pulmonary hyper- tension, right-to-left shunting and systemic hypoxemia. Subsequently, Murdock, ~3 Rowe, ~4 and Collins 15 made important contributions towards the recognition and treatment of pulmo- nary hypertension secondary to congenital dia- phragmatic hernia (CDH).

Profound systemic hypoxia results if pulmo- nary vascular resistance remains elevated post- natally (Fig 5). Pulmonary blood flow is dimin-

NPII in PERSISTENT PULMONARY HYPERTENSION

(%) pH

100 - - 7.6

90 - - 7.5

80 - - 7.4

70 - - 7 . 3

6 0 - - 7 , 2

5 0 - - 7 . 1

4 0 - - 7.0

30 - - 6 . 9

20 - - 6 .8

16 hrs

~5 hrs Qt

i i 4 8 lt2 lt6

NPit ~- 11 TIME (hours)

Fig 2. NPII calculation with PPHN.

2t0 .J 24

KRUMMEL ET AL

Table 4. Statistical Analysis. Mortal i ty Prediction

(A-a)Oo 2 >_ 600 torr for _> 12 hrs.

Sensitivity 0 .888

Specificity 0 .933

Accuracy of positive prediction 0 .960

Accuracy of negative prediction 0 .823

ished as prenatal pathways shunt venous blood right to left at atrial and ductal levels. Placental gas exchange is of course no longer available to provide oxygenation. Thus hypoxia results. Because the placenta is absent, the term "persis- tent fetal circulation (PFC)" is perhaps less descriptive than the term "persistent pulmonary hypertension of the newborn (PPHN)" sug- gested by Rudolph. 12

In these near-term infants, PPHN was uni- formly identified and presented the ususal pat- tern of hypoxia without hypercarbia. Respiratory acidosis did not occur: In fact, these infants were hyperventilated to enlist the purported pulmo- nary vasodilatory effects of alkalosis. In such a situation, NPII is not well suited to prediction of outcome since the area encompassed between the simultaneous plots of pH and Fio2 becomes small with induced alkalosis (Fig 2). Only as a preter- minal event did significant acidosis develop.

FETAL CIRCULATORY PATTERN

Fig 3. Normal fetal circulatory pattern.

MORTALITY PREDICTION IN ECMO CANDIDATES

NORMAL NEWBORN PERSISTENT PULMONARY HYPERTENSION

383

Fig 5, Circulatory f low patterns with persistent pulmo- nary hypertension of the newborn (PPHN).

Fig 4. Normal newborn circulatory pattern.

Previous application of (A-a)Do2 to mortality prediction in infants with CDH by Raphaely and Downes 9 has proven a relatively useful quantifi- cation of hypoxia over the years. Wider applica- tion to mortality prediction in neonates with PPHN by OrmazabaP ~ has demonstrated uni- form mortality when (A-a)D02 >_- 620 torr for a 12-hour period, and application of ECMO as a final resuscitative maneuver based on this mor- tality prediction has been reported? An unfortu- nate number of nonsurvivors have been excluded, however, prompting this reassessment of selec- tion criteria.

Accurate discrimination of survivors from nonsurvivors is necessary, in order to minimize the risk of ECMO, yet to allow early accurate

identification of the neonate certain to die despite best management, in order to maximize the benefits of ECMO. Such identification also permits meaningful comparison of outcomes with standard as well as newer methods of treat- ment. It appears that in these hypoxic newborns, a group of patients with 94% mortality can be correctly identified, by the presence of serial (A-a)Do2 >_ 600 torr over a 12-hour period. The accuracy for positive prediction is 0.960 and the accuracy for negative prediction is 0.823, answering respectively the questions: " I f an infant meets criteria, how likely is it to die?", and " I f an infant does not meet this criteria, how likely is it to live?". ~4 Serial (A-a)Do2 determi- nations appear to best identify the hypoxic infant with P P H N certain to die, and permit orderly application of ECMO.

REFERENCES

1. Bartlett RH, Gazzaniga AB, Fong SW, et al: Extracor- poreal circulation (ECMO) in neonatal respiratory failure. J Thorac. Cardiovasc. Surg. 74:826-835, 1977

2. Bartlett RH, Andrews AF, Toomasian JM, et al: Extra- corporeal membrane oxygenation (ECMO) for newborn res- piratory failure~45 cases. Surgery 92:425, 1982

3. Hardesty RL, Griffith BP, Debski RF, et al: Extracor-

poreal membrane oxygenation: Successful treatment of per- sistent fetal circulation following repair of congenital dia- phragmatic hernia. J Thorac Cardiovasc Surg 81:556-563, 1981

4. Krummel TM, Greenfield L J, Kirkpatrick BV, et al: Clinical use of an extracorporeal membrane oxygenator in

384 KRUMMEL ET AL

neonatal pulmonary failure. J Pediatr Surg 17:525-531, 1982

5. Krummel TM, Greenfield L J, Kirkpatrick BV, et al: The early evaluation of survivors after extracorporeal mem- brane oxygenation for neonatal pulmonary failure. J Pediatr Surg 1983 (in press)

6. O'Connor M J, Huxtable RF, Bartlett RH, et al: Neo- natal pulmonary insufficiency index: A method of newborn mortality prediction (submitted for publication)

7. Cimma R, Risemberg H, White J J, A simple objective system for early recognition of overwhelming neonatal respi- ratory distress. J Pediatr Surg 15:581-585, 1980

8. Dibbins AW, Weiner ES: Mortality from neonatal diaphragmatic hernia. J Pediatr Surg 9:653-662, 1974

9. Raphaely RC, Downs J J, Jr.: Congenital diaphrag- matic hernia: Prediction of survival. J Pediatr Surg 8:815- 823, 1973

10. Ormazabel M, Kirkpatrick B, Mueller D: Alteration of Alveolar-Arterial 02 gradient (A-aDO2) in response to tolazoline as a predictor of outcome in neonates with persis-

tent pulmonary hypertension (PPH). Pediatr Res 14:607, 1980

11. Gersony WM, Due GV, Sinclair JC: "PFC" syndrome (Persistence of fetal circulation). Circulation 39:11 I, 1969

12. Rudolph AM: High pulmonary vascular resistance after birth. Clin Ped 19:585-590, 1980

13. Murdock AI, Burrington JB, Swyer PR: Alveolar to arterial oxygen tension difference in newly born infants with congenital diaphragmatic herniation through the foramen of Bochdalek. Biol Neonate 17:161-172, 1971

14. Rowe MI, Uribe FL: Diaphragmatic hernia in the newborn infant: Blood gas and pH considerations. Surgery 70:758-761, 1971

15. Collins DL, Pomerance J J, Travis KW, et al: A new approach to congenital posterolateral diaphragmatic hernia. J Pediatr Surg 12:149-156, 1977

16. Ransohoff DF, Feinstein AR: Problems of spectrum and bias in evaluating the efficacy of diagnostic tests. New Engl J Med 299:926-930, 1978