Disease can profoundly impair lung function and may expose patients to increased risk for cell death.

 The measurement of arterial blood gases (ABGs) is an indispensable component of patient evaluation and follow-up in severe respiratory and renal diseases. Basic measurements obtained from an arterial blood sample may be insufficient, however, to give a complete picture of physiological impairment and to help clinicians determine appropriate treatment. Practical aspects of ABG analysis are covered elsewhere,1 but it is also important to understand derived ABG parameters and additional testing.

Hypoxemia is a major problem in respiratory medicine, both because it is very common and because its rapid evaluation and treatment are necessary to prevent irreversible organ damage. The direct measurement of ABGs is fundamental for clinical purposes. Recent gas analyzers have transformed Pao2 measurement into a simple, rapid procedure with an acceptable risk of error, but Pao2 measurement alone is inadequate for estimating the efficiency of gas exchange.2 Therefore, several indices derived from ABG measurements have been developed to improve the accuracy of estimates of gas-exchange efficiency. The ideal index is theoretically unaffected by changes in the fraction of inspired oxygen (FIo2) and does not require the sampling of mixed venous blood. The ideal alveolar-arterial difference in partial pressure of oxygen (PAo2–Pao2) can be used to distinguish hypoventilation from other causes of hypoxemia,3 but it cannot be used to differentiate among diffusion abnormality, ventilation-perfusion ratio (V/Q) inequalities, or shunt. In healthy young subjects, the PAo2–Pao2 is less than 10 mm Hg, and it increases to 10 to 20 mm Hg in the elderly.4 The tendency of PAo2–Pao2 to change as the FIo2 changes limits its value as an indicator of normal gas exchange.5 The Pao2/PAHo2 ratio was introduced in an attempt to develop a more stable index (unaffected by changing levels of FIo2), but it offered no significant advantages.5 This index can, however, be used to predict the FIo2 needed to attain a desired Pao2.6 The easiest index to calculate is the arterial inspired oxygen concentration ratio (Pao2/FIo2). It has been found to predict the shunt fraction reasonably well,5 and to be the most reliable index of gas exchange at FIo2 levels of 50% or less and Pao2 levels of 100 mm Hg or more.7

Physiological shunt is a very useful measure of the efficiency of arterial oxygenation and is independent of the shape of the oxygen-dissociation curve. Shunt requires a measurement of pulmonary arterial oxygen content (or simply saturation, for clinical purposes). It can be computed as oxygen content at the ideal point of the oxygen-carbon dioxide diagram minus arterial oxygen content, divided by the same ideal oxygen content minus mixed venous oxygen content. The problem is that measurement of mixed venous oxygen content requires blood to be sampled from the pulmonary artery. The normal physiological shunt is less than 5% and includes a small contribution (about 1%) from the extrapulmonary or obligatory shunt (anatomic shunt). The anatomic shunt is computed using the physiological shunt equation, but the measurements are performed while the subject is breathing 100% oxygen from a Douglas bag. An arterial sample is taken for Po2, Pco2, and hemoglobin after at least 15 minutes spent breathing to wash out the nitrogen from all gas-containing alveoli. This will eliminate V/Q inequalities as a cause of any PAo2–Pao2 gradient. Perfusion of airless lobes or lobules too large to be oxygenated by diffusion from neighboring normal lung behaves like an anatomic shunt. Measurement of the anatomic shunt is affected mainly by the failure to eliminate nitrogen completely from slowly ventilated acinar units. It follows that this method is unreliable in the presence of moderate-to-severe airflow obstruction. As an alternative, anatomic shunt can be estimated by injecting radiolabeled albumin particles large enough (more than 15 mm) to be trapped inside normal pulmonary microvascular channels. In anatomic shunting, a measurable fraction of albumin particles reaches the systemic circulation, so it can be measured with a gamma camera focused on high-flow organs such as the kidneys and brain. A good correlation has been found between the radioisotope and 100%-oxygen methods.8

F03a.jpg (15095 bytes)Figure 1. Distribution of the ventilation-perfusion ratio in a normal subject according to the 50 compartment model (MIGET).
o = ventilation
 •  = perfusion
Wagner PD, Lavaruso RB, Uhl RR, West JB. Continuous distributions of ventilation-perfusion ratios in normal subjects breathing air and 100% O2. J Clin Invest. 1974; 54: 54-68. Reprinted with permission.

Assessing V/Q inequalities is not easy due to the small volume (about 0.06 mL) of each acinus in relation to the resolution of the tools available (radionuclide techniques and flexible bronchoscopes used for direct gas sampling). The most sophisticated technique is the multiple inert gas elimination technique developed by Wagner et al.9 It measures the distribution of V/Q in a 50-compartment model of the lung (Figure 1). Such resolution constitutes a major improvement, but its complexity limits its use. Six inert gases with a wide range of solubilities are infused intravenously for 30 minutes, after which mixed venous, arterial, and mixed expired samples are taken and analyzed using gas chromatography or a mass spectrometer for the arterial retention and alveolar excretion ratio of each gas. The cardiac output and minute ventilation are also measured. With these data, a mathematical model can be developed in which V/Q distribution is fitted to 48 discrete values, plus one for shunt and one for alveolar dead space. Even if this 50-compartment approach has no links with the topographical distribution of real V/Q inequalities in single subjects, it has provided useful information on the V/Q ratio in different respiratory conditions, as well as on the pathological mechanisms leading to hypoxemia (Figure 2).10

F03b.jpg (20316 bytes)Figure 2. Distribution of the ventilation-perfusion ratio in two patients with chronic obstructive pulmonary disease.
    Upper panel: representative patient with predominant emphysema.     
    Lower panel: representative patient with predominant bronchitis.
Wagner PD, Dantzker R, Dueck JL, Clausen JL, West JB. Ventilation-perfusion inequality in chronic obstructive pulmonary disease. J Clin Invest. 1977;59:203-216.
Reprinted with permission.

Maldistribution of ventilation can be assessed in the clinical setting using five different techniques: multibreath nitrogen washout, phases III and IV of single-breath nitrogen expiration, radioactive gas and radiolabeled aerosol scans, frequency dependence of compliance, and volume difference in total lung capacity between single-breath and multi-breath or plethysmographic measurements. Maldistribution of ventilation is usually associated with airflow obstruction, the most common cause of V/Q mismatching and arterial hypoxemia. This condition is explained by slow gas mixing occurring during wash-in or washout techniques (long equilibration times). Multibreath nitrogen washout is the oldest (and probably the most sensitive) technique, but it is time-consuming. The single-breath nitrogen slope is easier to obtain (Figure 3). It is an analysis of the nitrogen content and volume of an exhalation following a vital-capacity inhalation of 100% oxygen. Four phases can be detected. In phase I, the anatomic dead space empties; phase II represents the transition in emptying from the bronchial and bronchiolar compartments to acinar/alveolar units; phase III reflects alveolar emptying, with a positive slope indicating uneven distribution of ventilation within and between lung regions; and phase IV reflects a change between regions in the pattern of lung emptying. This phase is caused by functional closure of airways in the dependent parts of the lung. The volume change from the onset of closure to the end of expiration is the closing volume, whereas the closing capacity is the closing volume plus residual volume. There are conditions (such as severe obesity) in which the closing volume exceeds the expiratory reserve volume. In this case, some airways will remain closed during the tidal breathing cycle. This can cause some degree of hypoxemia.11 Closing volume increases with age, probably due to declining elasticity in the aging lung.

F03c.jpg (15228 bytes)Figure 3. Closing volume measurement with the single breath nitrogen expiration technique. Appendini et al. 2005.

Radioactive gas measurements have been used to show variations in alveolar expansion, ventilation, and blood flow in the lung from nondependent to dependent regions.12,13 The differences between regions seen on ventilation scans underestimate the amount of inhomogeneity, compared with other methods. In the clinical setting, planar images of blood flow and ventilation are obtained using a gamma camera and interpreted using visual inspection. Radiolabeled aerosols are easier to handle than radioactive gases, but they overestimate regional differences because of limited diffusive mixing and by impaction and deposition of the aerosol in the central airways of patients. Radioaerosols of very small particle size are also available. Perfusion distribution is obtained through intravenous injection of radiolabeled albumin particles (15 to 60 µm in diameter) that are trapped in small pulmonary arteries in proportion to local blood flow. Diagnosis of pulmonary embolic disease is the main clinical application of ventilation and perfusion scans. Segmental defects in the perfusion scan in the presence of normal regional ventilation are diagnostic of pulmonary embolism. Concomitant defects of ventilation and perfusion distributed mainly in a subsegmental fashion usually are apparent in chronic obstructive pulmonary disease with high levels of parenchymal destruction. A normal perfusion scan rules out the diagnosis of pulmonary embolism. Ventilation and perfusion scans are also used for preoperative evaluation of the functional contribution of the lung to be resected.

Noninvasive Techniques
Noninvasive monitoring of arterial oxygenation was introduced to overcome the need for arterial puncture in situations requiring frequent determinations of Pao2. Pulse oximetry (Spo2) and transcutaneous measurement of Pao2 do not require arterial puncture. Carboxyhemoglobin and methemoglobin can cause oxygenation to be overestimated, but pulse oximetry is considered acceptably accurate at rest and during exercise.14 The strength of pulse oximetry lies in its ability to document saturation changes (from rest to exercise, from air to oxygen breathing, and for continuous monitoring at night). Another application is in situations where arterial sampling is difficult (for example, for patients at home). Pulse oximetry can be considered the most patient-friendly and user-friendly device available for noninvasive arterial oxygenation monitoring. Its main disadvantage is the fact that, due to the sigmoid shape of the oxygen dissociation curve, Spo2 is a less sensitive index than Pao2 of mild degrees of hypoxemia (Pao2 more than 60 mm Hg). Below this threshold, Spo2 becomes more sensitive.

In transcutaneous measurement of Pao2, the skin is heated to 40° to 42°C. Erythema and burns should be avoided by moving the electrode frequently. The method works better in neonates than in adults, since the epidermis is very thin in neonates. Individual differences in the anatomy and physiology of the dermis and epidermis are common in adults, so calibration with a simultaneous arterial sample is necessary. Transcutaneous measurement is well established for monitoring long-term trends in both Pao2 and Paco2.

The effects of gas-exchange impairment are less apparent on Paco2 than on Pao2 because the slope of the dissociation curve for carbon dioxide is steeper and more linear than that for oxygen. Therefore, carbon dioxide transport is more dependent on changes in ventilation. Hyperventilation is more effective in lowering Paco2 than in raising Pao2 because the sigmoid shape of the oxygen-dissociation curve limits further oxygen uptake at Pao2 levels of more than 90 mm Hg. This is why hypercapnia has been considered a hallmark of ventilatory pump failure rather than gas-exchange impairment.15

Hypercapnia has been defined as a Paco2 of more than 45 mm Hg.16 This value depends on altitude, however; at about 4.1 km above sea level, the normal Paco2 range is 28 to 32 mm Hg and a Paco2 of 40 mm Hg should be considered hypercapnic.

The first step in the evaluation and management of hypercapnia is to establish whether the rise in Paco2 is acute or chronic. The Henderson-Hasselbach equation and nomograms are the most accurate differentiation tools. Simpler rules are also available, though they are less precise.17,18

First, in acute hypercapnia, for each 10 mm Hg increase of Paco2, pH should decrease by 0.08 units from its normal baseline value (around 7.4) and the serum bicarbonate concentration should increase by 1 mEq/L. Second, in chronic hypercapnia, the same Paco2 increase should result in a pH decrease of only 0.03 units and in a serum bicarbonate increase of 3.5 mEq/L. Third, when acute hypercapnia is superimposed on chronic hypercapnia, pH and serum bicarbonate changes should be intermediate. Fourth, if pH and serum bicarbonate changes significantly differ from these values, a superimposed metabolic disturbance should be suspected.

A patient can develop hypercapnia either because he or she is unable to increase minute ventilation sufficiently or because there is an impairment of the ventilatory control of breathing. The two conditions can be differentiated by drawing an arterial blood sample during voluntary hyperventilation, although clinical evaluation is usually sufficient to guide diagnosis. Ventilation-perfusion inequality worsens hypercapnia in both conditions, since it greatly increases the amount of minute ventilation necessary to normalize the Paco2 level.

Acid-Base Balance
The amount of extracellular-fluid (ECF) and intracellular-fluid (ICF) hydrogen ions in the body can profoundly affect the rate of metabolic reactions by impairing enzyme activity.19 In humans, the acid-base balance is controlled by the renal and respiratory systems, which influence the ECF pH by changing bicarbonate and carbon dioxide concentrations. Renal and respiratory systems make up the bicarbonate-carbonic-acid buffer system. The Paco2 can be modified by a change in ventilation, and the bicarbonate level can be regulated by renal mechanisms. All other body buffer systems adjust accordingly to alterations in the bicarbonate and Paco2 pair; this relationship is known as the isohydric principle.

Variations of Paco2 alter the pH, rapidly affecting both ICF and ECF. Arterial Paco2 varies inversely with alveolar ventilation and varies directly with carbon dioxide production. Regulation of ventilation is the result of complex mechanisms linked to one another. Any change in arterial bicarbonate, Paco2, pH, Pao2, or stimuli from pulmonary mechanoreceptors changes minute ventilation (and, thus, Paco2 and pH). The respiratory response to a metabolic pH change follows peripheral chemoreceptor stimulation and provides the compensatory response to metabolic pH change.

Renal function alters the bicarbonate level, but the response is slow, and the maximum daily excretory capacity is reached only after 7 to 10 days.20 Bicarbonate has a rapid turnover: it may be generated by carbon dioxide or lost with carbon dioxide excretion.20,21

Renal and respiratory compensatory responses may be qualified as partial.22 Three parameters are necessary to describe an acid-base defect: pH, Paco2 (the respiratory component), bicarbonate (the metabolic component). The first two are measured directly, but there is no direct method for measuring bicarbonate concentration. Derived indices of standard bicarbonate, base excess, and buffer base have been developed in vitro to separate respiratory and metabolic bicarbonate components, but these indices do not accurately reflect the situation in vivo.23 In clinical practice, bicarbonate estimated using the Henderson-Hasselbach equation is considered sufficient (with pH and Paco2) for thorough interpretation of the acid-base disorders.23

Biochemical data (including ABG analysis, anion gap, and renal and hepatic serum profiles) should be taken into account. In mixed acid-base disorders, an acid-base diagram can be helpful.24 In this diagram, confidence intervals of pH and bicarbonate, in the form of confidence bands, are provided for each primary acid-base disorder and for the associated respiratory or renal compensation. In the absence of a diagram, simple rules to be used at the bedside have been developed.17,18,25 A primary metabolic acidosis is associated with a respiratory compensatory decrease in Paco2. Its numerical value is usually within 5 mm Hg of the two digits after the decimal point of the pH value, down to a pH of 7.15 to 7.1. A primary metabolic alkalosis is associated with a similar Paco2 change, increasing until the pH value reaches 7.55 to 7.6.

Metabolic acidosis results from a pathophysiological derangement generating excess noncarbonic acid or from abnormal gastrointestinal or renal loss of bicarbonate. ABG analysis usually shows a pH of less than 7.36, a Paco2 of more than 35 mm Hg, and a calculated bicarbonate level of less than 18 mmol/L. An increased anion gap may be present, indicating the accumulation of unmeasured acid anions.

Lactic acidosis may be classified as type A (seen during intense exercise, cardiac arrest, shock, hypoxia, and similar states), in which inadequate delivery of oxygen to the tissues generates lactate production in excess of the removal rate, or type B (seen in thiamine deficiency, diabetes, hepatic failure, renal failure, infection, pancreatitis, and the like), in which tissue hypoxia does not appear to play a major role.

Renal tubular acidosis is a disorder characterized by excess loss of urinary bicarbonate, a normal anion gap, and an elevated serum level of chloride. It is classified as proximal or distal, depending on the renal tubular site of the defect.

Metabolic alkalosis results from a pathophysiological derangement generating excess bicarbonate or from an abnormal loss of noncarbonic acid. ABG analysis shows a pH of more than 7.44, a Paco2 of more than 45 mm Hg, and a bicarbonate level of more than 32 mmol/L. The kidney has a large capacity to excrete bicarbonate, so maintenance of alkalosis requires an abnormal retention of bicarbonate, as seen in chronic renal failure, diminished functional ECF volume, severe potassium depletion, or mineralocorticoid excess with potassium depletion.

Lorenzo Appendini, MD, is assistant chief, pulmonary division; Antonio Patessio, MD, is assistant chief, pulmonary division, and head, respiratory pathophysiology laboratory; and Claudio F. Donner, MD, is chief, pulmonary disease division, Scientific Institute of Veruno, “Fondazione Salvatore Maugeri,” Italy. The authors thank Rosemary Allpress for her help in manuscript preparation.

1. Williams AJ. ABC of oxygen: assessing and interpreting arterial blood gases and acid-base balance. BMJ. 1998;317(7167):1213-16. Review.
2. West JB. Ventilation, blood flow and gas exchange. In: Murray JF, Nadel JA, eds. Murray and Nadel’s Textbook of Respiratory Medicine. Philadelphia: W.B. Saunders; 1994:51-89.
3. Snider GL. Interpretation of the arterial oxygen and carbon dioxide partial pressures: a simplified approach for bedside use. Chest. 1973;63(5):801-6.
4. Kanber GJ, King FW, Eschar YR, Sharp JT. The alveolar-arterial oxygen gradient in young and elderly men during air and oxygen breathing. Am Rev Respir Dis. 1968;97(3):376-81.
5. Zetterstrom H. Assessment of the efficiency of pulmonary oxygenation: the choice of oxygenation index. Acta Anaesthesiol Scand. 1988;32(7):579-84.
6. Hess D, Silage DA, Maxwell C. An arterial bloodgas interpretation program for handheld computers. Respir Care. 1984;29(7):756-9.
7. Gowda MS, Klocke RA. Variability of indices of hypoxemia in adult distress syndrome. Crit Care Med. 1997;25(1):41-5.
8. Whyte MKB, Peters AM, Hughes JMB, et al. Quantification of right to left shunt at rest and during exercise in patients with pulmonary arterio-
venous malformations. Thorax. 1992;47(10):790-6.
9. Wagner PD, Saltzman HA, West JB. Measurement of continuous distributions of ventilation-perfusion ratios: theory. J Appl Physiol. 1974;36(5):588-99.
10. West JB, Wagner PD. Ventilation-perfusion relationships. In: Crystal RG, West JB, eds. The Lung: Scientific Foundations. 2nd ed. Philadelphia: Lippincott-Raven; 1997:1693-1709.
11. Rea HH, Withy SJ, Seelye ER, Harris EA. The effects of posture in venous admixture and respiratory dead space in health. Am Rev Respir Dis. 1977;115(4):571-80.
12. Milic-Emili J. Topographical inequality of ventilation. In: Crystal RG, West JB, eds. The Lung: Scientific Foundations. 2nd ed. Philadelphia: Lippincott-Raven; 1997:1415-23.
13. West JB, Dollery CT, Naimark A. Distribution of blood flow in isolated lung: relation to vascular and alveolar pressure. J Appl Physiol. 1964;19:713-24.
14. Powers SK, Dodd S, Freeman J, et al. Accuracy of pulse oximetry to estimate HbO2 fraction of total Hb during exercise. J Appl Physiol. 1989;67(1):300-4.
15. Roussos C, Macklem PT. The respiratory muscles: medical progress. N Engl J Med. 1982;307(13):786-797.
16. Zakynthinos S, Roussos C. Hypercapnic respiratory failure. Respir Med. 1993;87(6):409-411.
17. Narins RG, Emmett M. Simple and mixed acid-base disorders: a practical approach. Medicine. 1980;59(3):161-187.
18. Aldrich TK, Prezant DJ. Indications for mechanical ventilation. In: Tobin MJ, ed. Principles and Practice of Mechanical Ventilation. New York: McGraw-Hill; 1994:155-189.
19. Relman AS. Metabolic consequences of acid-base disorders. Kidney. 1972;1:347-359.
20. Sartorius OW, Roemmelt JC, Pitts RF. The renal regulation of acid-base balance in man. The nature of renal compensations in ammonium chloride acidosis. J Clin Invest. 1949;28:423-439.
21. Pitts RF. Physiology of the Kidney and Body Fluids. 2nd ed. Chicago: Year Book; 1994.
22. Anderson OS, Astrup P, Bates RG, et al. Report of ad hoc committee on acid-base terminology. Ann NY Acad Sci. 1966;133:251-253.
23. Schwartz WB, Relman AS. A critique of parameters used in the evaluation of acid-base disorders. N Engl J Med. 1963;268:1382-1388.
24. Worthley LIG. A diagram to facilitate the understanding and therapy of mixed acid base disorders. Anaesth Intensive Care. 1976;4:245-253.
25. Worthley LIG. Hydrogen ion metabolism. Anaesth Intensive Care. 1977;5(4):347-360.