Assessment of Rest Hypoxemia: Is there a Place for the 100% Oxygen Test for Pathological Shunt Detection?

Background : The Hyperoxia Test (HT) has been proposed to quantify the shunt fraction but it is often requested as a first-line diagnostic tool, without clear guidelines on its place in the strategy for diagnosing hypoxemia. Objectives : To question the contribution of HT in managing patients with hypoxemia, taking into account comparison with morphological examinations. Methods : We retrospectively went through on patient files referred to the Pulmonary Functional Test (PFT) laboratory from January 2016 to December 2019. Clinical, PFTs and echocardiographic data were collected. Results : 53 cases were analyzed, HT showed a shunt above 7.5 % (HT+ group) in 42 (79%). Most of patients had a preexisting pulmonary disease. Rest PaO 2 and DLCOc/VA significantly differed between HT- and HT+ Eighteen of the HT+ patients had morphological examinations, five had an anatomic abnormality. Estimated shunt fraction was negatively strongly correlated with rest PaO 2 and DLCOc/VA. When compared to a predicted exploration strategy based on shunt value and PFTs abnormalities, the actual exploration strategy showed good concordance in HT- patients and some discrepancies in HT+ ones, but most of the time, there was a clinical reason to explain why a HT+ patients was no further investigated with echocardiography. Conclusion : HT appeared as an easy–to-do diagnostic tool in hypoxemic patients that could help to rule out pathologic shunt and to screen those eligible for more complex specialized exploration.


Introduction
Rest hypoxemia can be attributed to alteration of lung oxygenation function inducing abnormal distribution of the ventilation perfusion ratio (VA/Q) or disorders of the alveolarcapillary diffusion [1] leading to a shunt effect also called functional shunt. Nevertheless, when rest hypoxemia associates with normal or moderately altered Pulmonary Functional Tests (PFTs) results and normal imaging results of lung parenchyma, one should evoke an anatomic shunt. Even a shunt effect seems explained by abnormal PFTs, hypoxemia can lead to elevated lung arterial resistances, and thus right-sided pressures favoring Patent Foramen Ovale (PFO) thus anatomic and functional shunts can coexist and may interfere [2,3] on blood oxygenation, as possibly observed in COPD patients, Pulmonary Hypertension (PH), in sleep apnea disorders or in patients with unexplained refractory hypoxemia.
In healthy subjects a physiologic shunt may not exceed 5% of cardiac output at rest [4]. Higher values are considered as pathological and can be caused by PFO, other types of intracardiac defects, or less commonly to intra-pulmonary anatomical shunting via Pulmonary Arterio-Venous Malformations (PAM) [5] such as in the hepato-pulmonary syndrome [6] or the hereditary haemorrhagic telangiectasia disease. Despite intra-cardiac rightto-left shunting is one of the causes of hypoxemia [3], relationship between anatomical shunt, and hypoxemia remains unclear [7][8][9]. PFO closure can improve patients' outcome and quality of life [10][11][12]. The 100% oxygen inhalation test or Hyperoxia Test (HT) has been proposed since early 1960' to quantify the shunt fraction regardless its mechanism [13][14][15]. In 1968, S. T. Chiang proposed a formula to assess the shunt fraction after peripheral arteries blood gas analysis [16]. Visualization of PAM is ideally obtained using arteriography or injected computed tomography imaging. Concerning PFO, ultrasonic echocardiography is largely used. However, in hypoxemic patients, some limitations may be considered: contrast transesophageal echocardiography (cTEE) should not be tolerated well, contrast agents are contraindicated in case of severe emphysema [17] and trans-thoracic echocardiography (cTTE) can be thwarted because of low echogenicity as in obese or emphysematous patients [18]. Very few studies have focused on the contribution of the HT in hypoxemia diagnostic strategy. In hereditary hemorrhagic telangiectasia, Velthuis et al. have reported a low sensitivity of the HT in detecting abnormal shunt (>5%) as compared to contrast cTTE as HT led to 51% false negative results [19]. Because high resolution imaging ultrasonographs are not always available for rapid routine examination, the HT could still have a place for respirologists in charge for hypoxemic patients.
Here, we described our experience in a real-life retrospective descriptive study from a University-Hospital pulmonary function laboratory, trying to precise the contribution of the HT in hypoxemia diagnostic strategy.

Study Design
We investigated a retrospectively files of patients referred to the PFT laboratory of the University Hospital for a HT during January 2016 to December 2019. For this analysis, no additional procedure was added to standard of care. In the consultation ward, information about the analysis was available and the possibility was given to each patient to refuse the inclusion of self-medical data in the analysis. The collection and analysis of patients' data was approved by a local committee as per French legislation and is registered with the reference RGPD-APHM 20-251. Charts from patients of both gender in whom a HT was performed were included. In the absence of at least a spirometry test performed in the 7 days of the arterial blood gas samplings and in case of no confirmed hypoxemia,  The Netherlands). Diffusing capacity for the lungs was measured by the single breath method using carbon monoxide (DLCO) with helium dilution [20]. Diffusing capacity for carbon monoxide per unit of alveolar volume (DLCO/VA or KCO) and it's correction by hemoglobin concentration (DLCOc/VA) were evaluated. Values were expressed in their physical unit and compared to the reference CECA 93 equations' values [21]. The lung functional were established according to ATS/ERS definitions [22]. The spirometrybased GOLD severity of COPD was graded from 1 (mild) to 4 (very severe) [23]. The severity of restrictive pattern and altered diffusing capacity of the lung for carbon monoxide were graded according to ATS/ERS classifications [22]. Hypoxemia was defined as partial arterial pressure of oxygen (PaO 2 ) value below predicted [24].

Functional and Blood Gas Measurements
The HT was performed in the sitting position, the patient breathed 100% oxygen from a Douglas bag via a closely fitting mask. Arterial blood sample was collected after 30-min exposition.
The following Chiang's formula allowed to calculate the estimated % to consider abnormal shunt fraction. Thus shunt was estimated as pathologic (HT+ group when shunt ≥ 7.5 %) or physiologic (HTgroup when shunt < 7.5%).

Results and Shunt Value
For the purpose of this study, before any clinical chart review, the shunt value was confronted to the PFTs results. An exploration management strategy was then predicted according to the following algorithm that we established on the basis of our common clinical sense and experience ( Figure 1): Shunt value <7.5% : complementary anatomical explorations were considered as OPTIONAL (not required since hypoxemia may be not shuntrelated). Shunt between 7.5% and 20%:

Comparison of Predicted and Actual Managements
Each shunt/PFT picture was allocated to the REQUIRED or OPTIONAL complementary exploration strategy by a person blinded of the actual strategy. The "actual" exploration strategy, i.e.
that was really performed in the current practice after HT results, was then checked. Cases of discrepancies were then detailed.

Complementary Investigations
Intracardiac shunt was assessed by cTTE or cTEE. PAM was assessed using arteriography or injected computed tomography. PH was confirmed using a right heart catheterization. Data of Systolic

Statistical Analysis
We

Complementary Investigations in HT+ Patients
Of the 42 HT+ patients, 18 were further explored, none of the  (Table 3). In the 18 explored patients, FEV1 and FEV1 / FVC tended to be lower in those with anatomical defects than in those with functional shunt (Table   4).

Comparison of Predictive and Actual Management Strategies
Analysis between the predicted and the actual exploration

Discussion
This real-life study on current practice of HT in addition to PFTs showed that 1) most of the suspected cases (79%) of elevated shunt fraction were confirmed by HT at the retained threshold of 7.5% 2) DLCOc/VA was the most discriminant to decipher between elevated and normal shunt values 3) the patients in whom a PFO detection was plausibly indicated, 25% had no assessment for anatomical shunt detection. Easy to use in a PFTs laboratory, HT can be more rapidly scheduled than specialized ultrasonic examinations. In our center, HT was often requested in first-line as a screening test.
Patients were mainly COPD and DILD, some being planned for lung transplantation surgery. To decrease false positive results, the threshold is usually elevated above the physiologic value of 5% for the shunt. Here, we arbitrarily fixed at 7.5% the threshold.
The 100% concordance between predicted and actual exploration strategies in the HT-group suggests that HT was taken into account by physician in charge of the patient with high confidence to rule-out pathologic shunt. Between 7.5 and 20%, interpretation of a shunt value appeared more variable. In the 17 patients in whom complementary explorations were considered as OPTIONAL according to our algorithm, 3 were explored, indicating that when severe PFTs abnormalities are present, there was a good agreement between predicted and actual exploration strategies. The presence of a PFO in one of these cases did not influence his clinical management. It is plausible that PFO detection in a patient with shunt value <20%, was only a reflect of the high incidence of PFO in human. PFO is a remnant of the fetal circulation documented in 25% to 40% of general population [25,26]. The clinical management of some of our cases with shunt values > 20% would require special attention, only 9 of 14 being explored. Even considering that one 91-y old patient would not benefit incremental explorations and considering patients with severe emphysema and hypoxemia or unstable coronary disease as anechogenic on cTTE or unable to tolerate cTEE, it remained only 2 patients at risk for PFO who were not explored. Their respective shunt was 24 and 26.4%.
Our 33% rate of explored COPD patients who were confirmed to have a PFO agrees with that obtained in autopsy studies in the general population [25] or in patients with respiratory impairment [9]. Interestingly, hypoxia was not more profound in HT+ patients actually explored with cTTE or cTEE, than in non-explored ones suggesting that the severity of hypoxemia was not the main factor influencing further investigation demand. On the other hand, explored patients had better lung volumes, suggesting that the discrepancies between severity of hypoxemia and relatively good forced and total lung volumes was the motivation for further investigations. Recent study has shown that PFO in the COPD patient is as common as in the general population. However, when a PFO is present in a COPD patient, it is of more severe grade than for healthy subjects [9]. Here, FEV1 and FEV1/FVC tended to be lower in the anatomical shunt sub-group comforting previously published data in severe COPD patients. Both cTEE and cTTE allow arguing for the cardiac or extra-cardiac origin of anatomic shunting [27].
Apart from cardiac ultrasonic examinations, Contrast Transcranial Doppler (cTCD) is now considered the most sensitive and specific method for shunt detection [28][29][30]. As compared to cTEE and cTTE, cTCD fails to distinguish intra-pulmonary and intra-cardiac shunts and should be reserved at the screening step [28] cTCD was not available at the time of the study in our center and was not reported in this series.
Because most of the explored patients were previously known has having lung function alteration and/or lung disease, it is unclear what precisely decided the physician in charge to request for a shunt detection except in patients with normal or moderately altered PFTs. In one case, the sudden aggravation in oxygen requirement in a functionally stable COPD patient appeared the main reason. In two lung transplant candidates, the shunt assessment was required as part of the medical plan. HT cannot decipher between functional and anatomic shunt but offers the advantage of a precise quantification of the shunt even if not denied of false positive or false negative results. DLCOc/VA values was the only PFTs parameter that significantly differed between HT+ and HT-groups, and DLCOc/VA was significantly correlated to the shunt value. Our study did not clarify whether PFTs abnormalities or shunt value could identify patients with anatomical shunt, mainly PFO. Interestingly restrictive as well as obstructive defects were observed in patients presenting PFO.

Conclusion
The benefits of the HT at first-line appear clear when it allows ruling out a shunt in hypoxemic patients. When a pathological shunt is documented, some patients remain unexplored but most