In this study, we sought to clarify the reproducibility of VO2 at maximal effort (VO2peak) and VO2 at ventilatory threshold (VO2@VT), an analog for anaerobic threshold, in patients ill with ME/CFS. Three studies to date [13, 17, 18] have demonstrated an abnormal post-exertional response to exercise in ME/CFS, but they do not agree as to which physiological measures fail to respond normally in ME/CFS. Second, we wanted to find out how a compromised test-retest response to exercise would impact a standard classification of functional impairment based on VO2peak or VO2@VT. Classification described by Weber and Janicki  was initially devised to categorize functional impairment/exercise intolerance in patients with chronic cardiac failure, although it is useful for other patient populations in which impaired gas exchange (oxygen consumption, carbon dioxide production, minute ventilation) contributes to exercise intolerance and limits physical function.
The test-retest changes in VO2peak that we observed are consistent with decrements reported in the three previous studies of two-day CPET response in ME/CFS [13, 17, 18], although the magnitude of decrease in VO2peak varied among these studies. In the first report to quantify an abnormal post-exertional response to exercise in ME/CFS, VanNess et al.  assessed the contribution of VO2peak measured in six females with ME/CFS and six inactive female controls to discriminate between groups. An index of maximum effort (e.g., RER) was not reported with this initial pilot study. Results indicated that a VO2peak decrement in test 2 alone correctly identified 6 of 6 ME/CFS and 5 of 6 controls, for an overall classification accuracy of 91.7%. Based on their reported mean data, VO2peak decreased during test 2 by ~22% (P = .03), in contrast to a smaller test-retest decrease observed herein of 13.8% (P < 0.001). The more robust sample size in our study may have contributed to the smaller decrease in the test-retest measures of VO2peak; however, for both studies, the test-retest decrement is considerably greater than <6-7% variability reported consistently in healthy subjects [19, 20, 23] and various patient populations [25–27, 29, 35–38].
A more recent two-day CPET assessment of ME/CFS by the same group  included 51 females with ME/CFS and 10 healthy, inactive controls. This study included measures at ventilatory threshold in a discriminant function analysis. Similar to their earlier study, CPET measures distinguished 95.1% of ME/CFS patients from healthy controls, with a cross-validation accuracy of 90.2%. The primary and secondary discriminating variables in this study were: 1) work at ventilatory threshold intensity (decreased ~55%) and 2) work at maximal intensity (decreased ~7%), respectively. In contrast to their first study , VO2peak did not contribute to the ability to distinguish ME/CFS patients in this cohort. Further, univariate analysis of VO2peak revealed no significant difference between test 1 and test 2 for ME/CFS, which was within normal variation for VO2peak.
Our results also differ from those of Vermeulen et al. , who measured VO2peak in 15 females with ME/CFS and 15 healthy female controls who were comparable in age and BMI. While there was a 2.2% increase (P < 0.05) in VO2peak controls, they observed a ~6.3% decrease in VO2peak (P < 0.01) in ME/CFS patients which is comparable to normal test-retest variation in healthy subjects. It is possible that methodological differences between their study and that of VanNess et al.  and our study contributed to the smaller decrease in VO2peak in ME/CFS patients that they detected. The cycle test protocol used by Vermeulen et al.  was not described in detail and appeared to vary between subjects. Reproducibility of gas exchange measures in healthy and other patient populations relies on consistent testing methodology . Presumably, the protocol used for the same subject did not vary between tests, although that was not stated explicitly. Additionally, authors stated that maximum effort was assessed using RER, but the RER criterion (ie., RER ≥ 1.1) was not stated, and RER values were not reported. This is an important measure to indicate magnitude of effort, without which it is questionable whether patients gave maximal effort on both CPETs.
In addition to a 13.8% decrease in VO2peak in ME/CFS patients, we also observed decreases in maximal work (12.5%) and maximal heart rate (9 bpm). Likewise, Snell et al.  reported a decrease in maximal work of 7%. In repeat tests of leg extension strength and endurance, Paul et al.  also demonstrated a delayed recovery in ME/CFS work output with a greater decrease in quadriceps extension strength and endurance compared to controls following a 24 h repeat test. Conversely, Vermeulen et al.  reported no significant test-retest difference in maximal heart rate or work in ME/CFS subjects.
We observed a statistically significant test-retest decrease in maximal O2pulse of 8.8%, indicating compromised oxygen delivery in ME/CFS patients following induction of post-exertional malaise. O2pulse, a surrogate measure for stroke volume and arterio-venous oxygen content difference (a-vDO2), is a predictor of mortality in patients with cardiovascular disease . It is an important index of heart function  and may also be associated with the onset of exercise-induced ischemia [42, 43], but is also a stable and reproducible measure over time in young athletes  as well as adult non-athletes . Vermeulen et al.  found a non-significant decrease of ~5% in maximal O2pulse in ME/CFS patients . When this group later measured cardiac output and O2pulse during a single CPET in 178 ME/CFS patients, lower values were found in ME/CFS at VT and maximal intensities, but not at rest, compared to 11 sedentary controls. Additionally, they reported a lower arterio-venous oxygen content difference, determined non-invasively based on VO2 and cardiac output, and attributed these findings to lower O2 extraction by muscles during exercise in ME/CFS . While it is not known how alteration in oxygen delivery/utilization occurs during a subsequent CPET in ME/CFS patients, these results and others  also suggest that the decrease in maximal O2pulse may partly explain the concomitant reduction in maximal workload in ME/CFS that we observed.
Our data showed a substantial decrease of 15.8% in test-retest VO2 at VT. Large decreases in VO2 at VT were also reported by VanNess et al.  (~27%) and Snell et al.  (~11%). Although the test-retest decrease (7%) reported by Vermeulen et al.  was not statistically significant, there was a significant group by test interaction (P < 0.05) due to an increase in control subjects. In contrast, gas exchange variables and work at VT are reliable and reproducible in healthy subjects and athletes [21, 48], including test-retest differences of 1.5% for VO2 (r = .82-.97, Standard Error of Measurement (SE
) = 2.64 ml.kg.min-1), and 1.5% for cycle work (SE
= 4.5 W) or treadmill velocity (SE
= 10 m.min-1) (r = .95-.99). Oxygen consumption at VT in cardiac patients (Weber class A, B, C) is also stable and reproducible in multiple measures over months, albeit with somewhat more variability (CV = 9.2%) compared to healthy subjects .
Work measured at VT decreased 21.3% in our subjects as well as a remarkable 55% reported by Snell et al. . VanNess et al.  did not report work at VT, and Vermeulen et al.  found no significant difference in the univariate comparison of test-retest work at VT, but did find a significant group by test interaction (P < 0.05). O2 pulse at VT decreased significantly in our subjects (12.6%) and in Vermeulen et al.  (9%) and was not reported by VanNess et al.  or Snell et al. .
Changes in physiological measures indicate a substantial post-exertional decrement in performance at ventilatory threshold in ME/CFS 24 hours after an initial CPET. Ventilatory or anaerobic threshold intensity indicates the workload, heart rate and/or oxygen consumption at which anaerobic metabolism begins to predominate. Thus, after induction of post-exertional malaise, the threshold lowers at which anaerobic metabolism accelerates in ME/CFS. This causes premature anaerobiosis in ME/CFS patients after they have endured an earlier physical challenge, further reducing the ability to do work. It is therefore not surprising that Snell et al.  found that work at ventilatory threshold contributed most substantially to differentiate ME/CFS from healthy controls.
Use of a single CPET only to indicate functional impairment in ME/CFS is problematic. The results of this study, and the consensus of the three previous studies of test-retest CPETs in ME-CFS patients, provide strong evidence of impaired physiological responses to exercise. More specifically, the abnormal post-exertional responses to exercise in ME/CFS are marked by test-retest decreases in VO2 and work at both maximum and ventilatory threshold intensities. Data from a single CPET resulted in classification of 12 of 22 patients as having little or no impairment, and eight as having mild/moderate impairment. Such individuals would likely be prescribed graded exercise therapy (GET) to improve aerobic capacity. However, data from the second CPET in this and prior studies [13, 17, 18] indicate that aerobic energy-producing processes fail to respond normally to exercise stress in ME/CFS patients. Thus, incautiously applied GET is likely to result in exacerbation of fatigue and other symptoms of ME/CFS patients.
Little is understood about the anomalous post-exertional response to exercise in ME/CFS. We know that our data does not result from any methodological or equipment problems, because during the same time period the ME/CFS patients were being tested, we performed several repeat CPETs on healthy individuals, who demonstrated comparable or better consistency and reproducibility for VO2peak compared to published values [19–21, 23, 48]. The consistently high RER values during CPET 2 provide sound evidence that ME/CFS patients can, in fact, work to maximal effort in a repeat CPET. Values for maximal RER of 1.17 and 1.14 that were reported in this study would be taken as an indication of strong, maximal efforts if reported in healthy subjects and athletes [49, 50]. ME/CFS patients currently represent a unique class of ill patients who do not reproduce maximal CPET measures, unlike individuals with cardiovascular disease [27, 30] lung disease , end-stage renal disease , pulmonary arterial hypertension  and cystic fibrosis .
A limitation of this study should be addressed in follow-up research. Together with the three previous studies of the two-day CPET protocol [13, 17, 18], the collective results demonstrate consistently abnormal CPET results in ME/CFS during test 2. However, the variation in abnormal CPET responses among these studies was not clarified in the present study and requires a larger sample size with robust statistical power.
Subsequent research should strive to address the following questions regarding post-exertional fatigue in ME/CFS. Inclusion of additional males in subsequent research should allow us to ascertain whether there are sex differences in response to the two-day CPET protocol. A large sample size will be needed to determine whether we can sub-classify ME/CFS patients based on differential responses to the two-day CPET protocol at maximal and ventilatory threshold intensities. With additional participants, it would be possible to identify clinically relevant exercise measurement cutpoints and odds ratios for use by practitioners in the diagnosis and treatment of those with ME/CFS. Physical activity prior to and following the two-day CPET should be quantified to correlate changes with the decrement measured during testing.