James E. Hansen, MD, FCCP; Gaye Ulubay, MD; Bing Fai Chow, MD; Xing-Guo Sun, MD; and Karlman Wasserman, PhD, MD, FCCP

Transcription

1 CHEST Original Research EXERCISE TESTING Mixed-Expired and End-Tidal CO 2 Distinguish Between Ventilation and Perfusion Defects During Exercise Testing in Patients With Lung and Heart Diseases* James E. Hansen, MD, FCCP; Gaye Ulubay, MD; Bing Fai Chow, MD; Xing-Guo Sun, MD; and Karlman Wasserman, PhD, MD, FCCP Background: Mismatching of ventilation to perfusion is found in patients with, left ventricular failure (), and pulmonary vascular diseases. Such mismatching may be due to ventilation or perfusion defects or both. Our primary hypothesis was that pressures of mixedexpired CO 2 pressure ( ), end-tidal PCO 2 pressure (PETCO 2 ), and their ratios would differ between groups during exercise testing, depending on whether the ventilation/perfusion (V /Q ) abnormality was dominantly caused by airways or perfusion defects. Methods: We administered incremental cycle ergometry tests to normal subjects and three groups of patients, each group with uncomplicated,, or primary pulmonary arterial hypertension (). We compared,petco 2, and their ratios at rest, unloaded pedaling, anaerobic threshold, and peak exercise. Results: Although each patient group had mean peak O 2 uptake of approximately 50% of predicted normal, the levels and patterns of change for each group for,petco 2, and their ratios were surprisingly distinctive. As hypothesized, the group always had markedly lower /PETCO 2 ratios than all other groups (p < 0.001). In addition, patients with had slightly lower /PETCO 2 ratios at heavy exercise than normal subjects (p < 0.05). At all times, except for group PETCO 2 at peak exercise, each group had significantly lower PETCO 2 and than normal subjects (p < 0.001). In patients with, the PETCO 2 decline with exercise was distinctive. Conclusions: The levels and changes in,petco 2, and their ratios during cardiopulmonary exercise testing are distinctive and explained by the differing pathophysiologies of V /Q mismatching in these disorders. (CHEST 07; 132: ) Key words: congestive heart failure; ; exercise test; pulmonary artery hypertension; pulmonary circulation; ventilation/perfusion ratio Abbreviations: AT anaerobic threshold; Dlco diffusion capacity of the lung for carbon monoxide; left ventricular failure; primary pulmonary arterial hypertension; Peco 2 mixed-expired CO 2 pressure; Petco 2 end-tidal CO 2 pressure; V co 2 carbon dioxide output; Vd/Vt dead space/tidal volume; V e expired ventilation; V o 2 oxygen uptake; V /Q ventilation/perfusion The mismatching of ventilation to perfusion is a common physiologic abnormality that accounts for much of the increased ventilatory response to exercise in a number of diseases including, pulmonary vascular disease leading to primary pulmonary arterial hypertension (), and left ventricular failure (). 1 The mechanisms of these disorders could be due to uneven distribution of ventilation or perfusion, or both. Having found cardiopulmonary exercise testing to be helpful in differential diagnosis and understanding the pathophysiologies of these disorders, 2 we addressed how to ascertain which mechanism might be dominant, particularly when two disorders such as and might coexist. Additionally, differentiation between and pulmonary vascular disease may CHEST / 132 / 3/ SEPTEMBER,

2 *From the Los Angeles Biomedical Research Institute (Drs. Hansen, Sun, and Wasserman), Harbor-UCLA Medical Center, Respiratory and Critical Care Division of Physiology and Medicine, Department of Medicine, University of California at Los Angeles, Torrance, CA; Department of Chest Diseases (Dr. Ulubay), Faculty of Medicine, Baskent University, Ankara, Turkey; and Department of Respiratory Medicine (Dr. Chow), Ruttonjee Hospital, Hong Kong, SAR. This work was performed at Los Angeles Biomedical Research Institute, Harbor-UCLA Medical Center. James E. Hansen has received consultation fees from MET-Test, LLC, and salary support from St. Jude Medical. Dr. Ulubay received support from the Turkish Scientific Research Institute (Tubitak). Dr. Chow reports no conflict of interest. Dr. Sun has received salary support from St. Jude Medical. Dr. Wasserman has received consultation fees from MET-Test, LLC, and salary support from St. Jude Medical. Manuscript received March 9, 07; revision accepted May 10, 07. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( org/misc/reprints.shtml). Correspondence to: James E. Hansen, MD, FCCP, Harbor-UCLA Medical Center, 1000 W. Carson St, RB-2, Box 405, Torrance CA 90509; DOI: /chest be difficult, since low peak O 2 uptake (V o 2 ), low anaerobic threshold (AT), low end-tidal Pco 2 (Petco 2 ), and a high ratio of expired ventilation (V e) to CO 2 output (V co 2 ) are common to both. 2 In an ideal lung, where ventilation is uniformly and precisely matched to perfusion, gas exchange is highly efficient. However, because the physiologic dead space/tidal volume (Vd/Vt) ratio is never zero, 3 Peco 2 is always lower than Petco 2. With poorer matching, Vd/Vt increases, whether mismatching is due to uneven perfusion, uneven ventilation, or a mixture of uneven perfusion and uneven ventilation, causing an even lower Peco 2. Thus, Peco 2 is reduced under all conditions of uneven ventilation/ perfusion (V /Q ), including,, and. The degree of dilution of Peco 2 relative to Petco 2 is more severe in airways disease (eg, ), where the increase in Vd/Vt is due to nonuniform emptying of the airways, with the longest time-constant low V /Q lung units contributing to the end-tidal gas. Contrariwise, cardiac or pulmonary vascular disorders have relatively uniform emptying of airways but nonuniform perfusion accounting for the increase in Vd/Vt. We therefore hypothesized that in airways diseases (eg, ), with a larger proportion of long timeconstant, low V /Q airways contributing to the Petco 2, the discrepancies between Peco 2 and Petco 2 would be greatest and the Peco 2 /Petco 2 ratios lowest. We also postulated that in patients with both and with relatively normal airways, both Petco 2 and Peco 2 would be dilute due to uneven or hypoperfusion of ventilated airspaces but with relatively normal Peco 2 /Petco 2 ratios. Possibly, the patterns of Peco 2,Petco 2, and their ratio in and patients would be somewhat distinctive. The objective of this study was to determine if Peco 2 and Petco 2, and their ratios, all noninvasive measures, might distinguish between abnormal V /Q due to primary airway disorders from those due to primary perfusion disorders during increasing work rate exercise in patients with relatively uncomplicated diagnoses of,, or. Subjects Materials and Methods The study was approved by our institutional review board. From our recent files, the authors screened and selected cardiopulmonary exercise test results obtained with informed consent from adults in each of the following four groups: normal,,, and. All normal subjects were considered healthy. All patients had FEV 1 /FVC ratios 50% and low or reduced gas transfer index; none had asthma. All patients were New York Heart Association class III or IV left ventricular systolic dysfunction heart failure. All patients had elevated pulmonary artery pressures and normal wedge pressures on right-heart catheterization. No patients had obstructive or interstitial lung disease or congenital heart disease. The mean values of peak V o 2 averaged 99% of predicted 2 for the normal group, and 45 to 50% of predicted for the three disease groups (Table 1). Procedures All tests were performed on a cycle ergometer with 3 min of rest, 3 min of unloaded cycling, and uniform increases in resistance of 5 to W/min for the patients and 10 to W/min for the normal subjects to maximal tolerance. The rates of increasing work depended on the estimated exercise capability of the subject. Gas exchange was measured using breath-by-breath instrumentation (Medical Graphics; St. Paul, MN; or Sensor- Medics; Anaheim, CA) with 50- to 60-mL mouthpiece dead space. 2 Calculations All values were obtained from continuous 10-s averages. Each AT was determined by the V-slope method, 4 with confirmation by viewing supporting data reflecting the development of a lactic acidosis. Peco 2 values were calculated from recordings of V co 2 and V e after subtraction of mouthpiece dead space ventilation. Although the V e/v co 2 ratio is commonly used to evaluate ventilatory inefficiency, in this study we substitute Peco 2, which is easily calculated by dividing 863 by the V e/v co 2 ratio, thus allowing direct comparison between Peco 2 and Petco 2.Petco 2 and Peco 2 values were averaged for the last minute of rest, last minute of unloaded cycling, 30 s to 1 min after the AT was reached, and for 30 s at peak exercise. Statistical Analysis Mean, SD, and SEM of values of Peco 2, Petco 2, and Peco 2 /Petco 2 were calculated. The significance of these three 978 Original Research

3 Table 1 Characteristics of Normal Subjects and Patients With,, and * Characteristics Normal Participants, No. Age, yr Male/female gender, No. 19/6 13/12 /5 2/23 Height, cm Weight, kg Peak V o 2, L/min Peak V o 2, % predicted Peak V co 2, L/min Peak V e, L/min FEV 1,L VC, L FEV 1 /VC, % Maximum voluntary ventilation, direct, L/min, Breathing reserve, L/min Total lung capacity, L Total lung capacity, % predicted Residual volume/total lung capacity, % predicted Dlco, U Dlco, % predicted *Data are presented as mean SD unless otherwise indicated. significantly different from at p significantly different from at p values between disease groups at each time period was assessed by analysis of variance, with critical differences between groups assessed by Tukey tests, 5 and between time periods by Scheffe tests, with p 0.05 considered significant. Pearson correlation coefficients were calculated for each group and each time period to ascertain if within-group Pco 2 values were significantly correlated with percentage of predicted peak V o 2 values, percentage of predicted gas transfer index (diffusion capacity of the lung for carbon monoxide (Dlco), and FEV 1 /VC when such data were available.. Results Demographics of the subjects are noted in Table 1. Although age and sex distributions were dissimilar, generally reflecting age and sex distributions seen in the respective disease states, disease severities (percentage of predicted peak V o 2 ), peak V e, and peak V co 2 for the three disease groups were not significantly different. However, FEV 1 /VC, Dlco, and breathing reserve were significantly lower and residual volume/ total lung capacity significantly higher in the group than the group. Figure 1, top, A, shows Petco 2 values for each group at four activity levels. Each group differed from every other at all activity levels (p 0.001), except for the vs groups at rest and the normal vs groups at peak exercise (p 0.05). The Petco 2 of each group changed significantly with each increasing level of activity, except for patients, in which Petco 2 did not change significantly from AT to peak exercise, attributable to their inability to increase ventilation further. Distinctly, Petco 2 decreased in the group as exercise intensity increased. In all other groups, Petco 2 rose significantly as exercise intensity increased to the AT. Figure 1, center, B, shows Peco 2 values at similar times as Figure 1, top, A. All patient groups had significantly lower Peco 2 than normal subjects at all activity levels (p 0.001). Peco 2 values rose in all groups from rest to the AT as activity intensity increased, albeit only 1 mm Hg in. Except for the group, Peco 2 fell significantly at peak exercise, attributable to the acidemia-induced hyperventilation and lower mean alveolar Pco 2. The only other significant differences were that Peco 2 levels were lower for patients than patients at rest, while Peco 2 levels for patients were lower than in patients at unloaded cycling and AT, and lower than in patients at peak exercise. Figure 1, bottom, C, shows the significantly reduced Peco 2 /Petco 2 ratios at all levels of activity in the group (p 0.001), confirming our major hypothesis. Consistently, the Peco 2 /Petco 2 ratios trended lower than in normal subjects for both and, with lower than. was significantly lower than in normal subjects only at AT and peak exercise (p 0.05). For each group, the ratios increased significantly with each increase in activity up to the AT, indicating universal improvement in V /Q with increasing exercise. Figure 2 integrates Peco 2 and Petco 2 and their ratios (diagonal dashed lines) for all four groups at all four activity levels progressing from rest (open sym- CHEST / 132 / 3/ SEPTEMBER,

4 PETCO 2 /PETCO 2 (Ratio) A B C N vs C *** *** *** NS N vs L *** *** *** *** N vs P *** *** *** *** C vs L NS *** *** *** C vs P *** *** *** *** L vs P * *** *** *** N vs C *** *** *** *** N vs L *** *** *** *** N vs P *** *** *** *** C vs L * NS NS NS C vs P NS NS NS ** L vs P NS * ** NS N vs C *** *** *** *** N vs L NS NS * * N vs P NS NS NS NS C vs L *** *** *** *** C vs P *** *** *** *** L vs P NS NS NS NS Rest Unloaded AT Peak Activity Figure 1. Values for normal subjects ( or N), (or C), (or L), and (or P) during incremental cycle ergometry tests at rest (open symbols), unloaded cycling, AT, and peak exercise (closed symbols). Values are mean SEM. Top, A: Petco 2 ; center, B: Peco 2 ; and bottom, C: Peco 2 / Petco 2. Statistically significant differences between groups are shown as *, **, and *** for p 0.05, p 0.01, and p 0.001, respectively, below value symbols. NS not significant Open=Rest /PETCO PETCO 2 Figure 2. Mean SEM Pco 2 values in normal,,, and groups during incremental cycle ergometry at rest (open symbols), progressing to unloaded cycling, AT, and peak exercise. Diagonal lines depict the Peco 2 /Petco 2 ratios from 0.8 to 0.5. Each group has a distinctive placement and distinctive activity pattern. See Figure 1 legend for expansion of abbreviation. bols) to solid symbols at unloaded cycling, AT, and peak exercise. In all groups, Peco 2 /Petco 2 ratio improved with exercise. Each group is distinctive. The normal group has the highest Peco 2,Petco 2, and Peco 2 /Petco 2 ratios, with the greatest Petco 2 changes between stages of activity. The group not only has the lowest Peco 2 /Petco 2 ratios, but also has the least change in Peco 2 and Petco 2 from AT to peak exercise. The group is distinctive with a progressive decline in Petco 2 from rest to peak. The group has Petco 2 values between the and groups, and lower Peco 2 / Petco 2 ratios than normal subjects at heavy exercise. To ascertain whether the differences between the and groups might be due predominantly to disease severity, the changes between more and less severe disease (as assessed by percentage of predicted peak V o 2 ) are shown in Figure 3. Within each disease group, the distinctive pattern of change remained, but in individual tests these patterns were not universally found. Within-group correlations are shown in Table 2. For, the only significant correlations with percentage of predicted peak V o 2 were vs Peco 2 / Petco 2 ratios at peak exercise, (ie, fewer long time-constant airspaces in less-limited patients), and with FEV 1 /VC vs Petco 2 at AT and peak exercise, (ie, higher Petco 2 with worse obstruction). For, Peco 2 always correlated positively with percentage of predicted peak V o 2 (ie, less ill had better gas-exchange efficiency), and Peco 2 /Petco 2 ratios Original Research

5 Open = Rest /PETCO Higher Lower Half Half PETCO 2 Figure 3. Mean Pco 2 values in subgroups of or from rest (open symbols) progressing to unloaded cycling, AT, and peak exercise (closed symbols). Although the distributions within each of the two -patient groups are bell shaped, the groups have each been classified into less severe (higher half) and more severe (lower half) subgroups based on their percentage of predicted peak V o 2 values. correlated positively with percentage of predicted Dlco (ie, apparently better V /Q matching occurred with better-perfused pulmonary vascular beds during exercise). For, there was a tendency for higher exercise Petco 2 values with higher percentage of predicted V o 2 ; the only significant correlation was negative between percentage of predicted peak V o 2 and Peco 2 /Petco 2 ratio at unloaded cycling, perhaps reflecting the trend to lower Petco 2 with more severe failure. Discussion Discerning the dominant disorder in patients with dyspnea can sometimes be difficult. As previously found in several studies, V /Q inequalities, increased Vd/Vt, and/or decreased Peco 2 and Petco 2 occur at rest and during exercise in, 6 10, 11 and The major hypothesis and finding of this study was that V /Q mismatch caused by airway defects could be differentiated from that of perfusion defects using noninvasive measures of Peco 2 and Petco 2. With mismatch of ventilation, Peco 2 would be dilute (ventilation of relatively short time-constant lung units contributing most of the V e) as related to Petco 2 (long time-constant lung units with low V /Q high Pco 2 lung units contributing least to the V e), and Peco 2 /Petco 2 ratio would be reduced. We also hypothesized and found that V /Q mismatch due to reduced or maldistributed pulmo nary blood flow, without airway defects, can reduce both Peco 2 and Petco 2, resulting in nearnormal Peco 2 /Petco 2 ratios. While other methods exist for determining or differentiating V /Q mismatch, they are generally invasive or require isotopic techniques. The method described here is noninvasive and is merely an analysis of the pressure and distribution of CO 2 in the exhaled breath. Thus, it is quite inexpensive and safe. Disease states were selected for study of the changes in Peco 2,Petco 2, and their ratio to distinguish between V /Q mismatch primarily due to perfusion abnormalities and V /Q mismatch primarily due to ventilation abnormalities. Each patient group differed from normal subjects and from each other even more than anticipated. First, the presence of delayed emptying of long time-constant airspaces may be seen in the Y-axis of Figure 1, bottom, C, and by the diagonal lines of the Peco 2 /Petco 2 ratios in Figure 2. The ratios are markedly impaired in (approximately 0.6 during exercise), mildly impaired in (approximately 0.7 during exercise), minimally impaired in (approximately 0.7 to 0.75 during exercise), and best in normal subjects (approximately 0.75 during exercise). Secondly, the inability of patients to force a decline in Petco 2 above the AT (Fig 1, top, A, 2) reflects their difficulty in responding to lactic acidemia by increasing ventilation. Third, the differences in Petco 2 (Fig 1, top, A, 2) and Peco 2 (Fig 1, center, B, 2) between the three patient groups at all intensities of exercise were substantial, considering that the mean percentage of predicted peak V o 2 was not different between these groups. The declining CO 2 values may reflect the progressively worse lung perfusion in,, and groups. As is not exclusively an airways disease (although FEV 1 correlates with Petco 2 within the group), the lower Dlco in is due both to maldistribution of ventilation and capillary destruction, while in and the lower Dlco is exclusively due to poor capillary perfusion. Fourth, the decreased Petco 2 with exercise in patients (Fig 2, 3) reflects their inability to adequately increase pulmonary perfusion with exercise, while all other groups can do so. Lastly, within each group, the absolute levels of Peco 2 and Petco 2 often correlate (Table 2) with apparent disease severity. The question has been raised as to whether the low Petco 2 and Peco 2 in and might be due to an exaggerated chemoreflex leading to increased V e, rather than greater pulmonary dead space ventilation due to perfusion defects In this study, we did not have Paco 2 or ph measurements to document the increase in Vd/Vt. However, CHEST / 132 / 3/ SEPTEMBER,

6 Table 2 Correlation Coefficients of Peak V O2,DLCO, and FEV 1 /Vital Capacity With,PETCO 2, and /PETCO 2 * Variables Rest Unloaded AT Peak Percentage of predicted peak V o 2 vs Petco 2 Normal Percentage of predicted peak V o 2 vs Peco 2 Normal Percentage of predicted peak V o 2 vs Peco 2 /Petco 2 Normal Percentage of predicted Dlco vs Peco 2 /Petco FEV 1 /vital capacity vs Petco *No other significant correlations with Dlco or FEV 1 /vital capacity. p p prior investigations with multiple arterial blood gas and ph measurements in patients with both chronic heart failure 22 and Eisenmenger disease 23 demonstrated not only an increased Vd/Vt, but also tight regulation of arterial ph and an exercise lactateinduced acidemia rather than alkalemia, as would be present if the primary cause of a lower Paco 2 (plus Peco 2 and Petco 2 ) were chemoreflex-induced hyperventilation. This prior study 22 concluded that regional hypoperfusion (of the lung) typifies gas exchange in heart failure. In the patients in the current study (Table 2), the degree of hypoperfusion as measured by the decrease in percentage of predicted Dlco correlated strongly with the decrease in Peco 2 /Petco 2 ratio (ie, an expected worsening Vd/ Vt). Further, percentage of predicted peak V o 2 correlated well with exercise Peco 2 and Petco 2 in patients. Clinically, the CO 2 exchange differences between the and groups are not always significant, but patients are frequently hypoxemic (with shunts or rapid blood flow through the pulmonary capillaries), while patients able to exercise (slow flow through the pulmonary capillaries) are rarely hypoxemic. The unique tendency of Petco 2 to decline with increasing exercise in most patients may be of some benefit in differentiating the cause of dyspnea when the etiology is uncertain. Despite other sources of information available to the physician, there are times in individual patients when the distinction between heart, other vascular, or lung disease, or deciding which is predominant can be challenging. Because exercise testing with gas exchange measurements is noninvasive, safe, and inexpensive, we find it very helpful not only in quantifying overall deficits but also in identifying the dominant V /Q disorder. One of the goals confronting practitioners today is the identification of individuals early in their disease states with safe and inexpensive tests in the hope that interventions at that time may be most salutary. Understanding the pathophysiology of the patterns seen in dyspneic patients should be helpful. References 1 Wasserman K, Hansen JE, Sue DY, et al. Principles of exercise testing and interpretation. 1st ed. Philadelphia, PA: Lea and Febiger, 1987; Wasserman K, Hansen JE, Sue DY, et al. Principles of exercise testing and interpretation. 4th ed. Philadelphia, PA: Lippincott, Williams and Wilkins, 05 3 Comroe JH, Forster RE, DuBois AB, et al. The lung: clinical physiology and pulmonary function tests. 2nd ed. Chicago, IL: Yearbook, Sue DY, Wasserman K, Moricca RB, et al. Metabolic acidosis during exercise in patients with chronic obstructive pulmonary disease. Chest 1988; 94: Dixon WJ, Massey FJ Jr. Introduction to statistical analysis. 3rd ed. New York, NY: McGraw-Hill, Wagner PD, Dantzker DR, Dueck R, et al. Ventilation- 982 Original Research

7 perfusion inequality in chronic obstructive pulmonary disease. J Clin Invest 1977; 59: Brown HV, Wasserman K. Exercise performance in chronic obstructive pulmonary disease. Med Clin North Am 1981; 65: Nery LE, Wasserman K, French W, et al. Contrasting cardiovascular and respiratory responses to exercise in mitral valve and obstructive pulmonary diseases. Chest 1983; 83: Liu Z, Vargas F, Stansbury D, et al. Comparison of end-tidal arterial Pco 2 gradient during exercise in normal subjects and in patients with severe. Chest 1995; 107: Palange P, Forte S, Onorati P, et al. Ventilatory and metabolic adaptations to walking and cycling in patients with. J Appl Physiol 00; 88: Rubin SA, Brown HV. Ventilation and gas exchange during exercise in severe chronic heart failure. Am Rev Respir Dis 1984; 129:S63 S64 12 Sullivan MJ, Higginbotham MB, Cobb FR. Increased exercise ventilation in patients with chronic heart failure: intact ventilatory control despite hemodynamic and pulmonary abnormalities. Circulation 1988; 77: Al-Rawas OA, Carter R, Richens D, et al. Ventilatory and gas exchange abnormalities on exercise in chronic heart failure. Eur Respir J 1995; 8: Wasserman K, Zhang YY, Riley MS. Ventilation during exercise in chronic heart failure. Basic Res Cardiol 1996; 91(suppl 1):1 11 Wensel R, Georgiadou P, Francis DP, et al. Differential contribution of dead space ventilation and low arterial Pco 2 to exercise hypernea in patients with chronic heart failure secondary to ischemic or idiopathic dilated cardiomyopathy. Am J Cardiol 04; 93: D Allonzo GE, Gianotti LA, Pohil RL, et al. Comparison of progressive exercise performance of normal subjects and patients with primary pulmonary hypertension. Chest 1987; 92: Reybrouck T, Mertens L, Schulze-Neick I, et al. Ventilatory inefficiency for carbon dioxide during exercise in patients with pulmonary hypertension. Clin Physiol 1998; 18: Yasunobu YY, Oudiz RJ, Sun XG, et al. End-tidal Pco 2 abnormality and exercise limitation in patients with primary pulmonary hypertension. Chest 05; 127: Ponikowski P, Chua TP, Anker SD, et al. Peripheral chemoreceptor hypersensitivity: an ominous sign in patients with chronic heart failure. Circulation 01; 104: Sun SY, Wang W, Zucker IH, et al. Enhanced activity of carotid body chemoreceptors in rabbits with heart failure: role of nitric oxide. J Appl Physiol 1999; 86: Sun SY, Wang W, Zucker IH, et al. Enhanced peripheral chemoreflex function in conscious rabbits with pacing-induced heart failure. J Appl Physiol 1999; 86: Wasserman K, Zhang YY, Gitt A, et al. Lung function and exercise gas exchange in chronic heart failure. Circulation 1997; 96: Sietsema KE, Cooper DM, Perloff SK, et al. Control of ventilation during exercise in patients with central venous-tosystemic arterial shunts. J Appl Physiol 1988; 64: CHEST / 132 / 3/ SEPTEMBER,