Timing Assessment of Response to Fluid Challenge in Patients with Septic Shock

Objectives: Fluid challenge (FC) is most commonly used for uid responsiveness (FR) evaluation, with a wide divergence in assessment time choices. Therefore, we aimed to explore the optimal assessment time for FC in patients with septic shock. Methods: A prospective cohort study was conducted. Septic shock patients who had experienced initial resuscitation and required an FC with 500 mL 4% gelatin or normal saline (NS) over 5-10 min were included. FR was dened by an increase in cardiac index (CI) >10%. FR and other predened variables were recorded at baseline (T b ), immediately (T 0 ), and at 10 (T 1 ), 30 (T 2 ), 45 (T 3 ), 60 (T 4 ), 90 (T 5 ), and 120 (T 6 ) min after FC. The incidence of FR and hemodynamic variables at predened time points were recorded. Data were analyzed by repeated measures of analysis of variance. Results: 63 patients were enrolled, with 43 in the gelatin group and 20 in the NS group. Among the 45/63 (71%) responders, 31 were responded at T 0 (ER), while 14 responded at T 1 or later (LR). The proportion of NR, ER and LR was comparable between gelatin and NS groups. After FC, the time course of FR status was slightly different between gelatin and NS groups. In the gelatin group, FC induced most responders (69%, 31/45) and frequency of CI maximum (35%, 11/31) at T 2 and sustained a positive FR status until T 4 ; while in the NS group, FC induced most responders (55%, 11/20) and frequency of CI maximum (64%, 9/14) at T 1 , and sustained FR status until T 1 . Conclusions: Different time courses of FR were found between gelatin and NS group patients undergoing FC. Thus, when NS is used, FR should be performed within 10 min, while it is better to extend the assessment time to 30 min after FC when gelatin is used. and oxygen metabolism variables for analysis at baseline (T b ), immediately (T 0 ), and at 10 (T 1 ), 30 (T 2 ), 45 (T 3 ), 60 (T 4 ), 90 (T 5 ) and 120 (T 6 ) min after FC. Hemodynamic variables included mean arterial pressure (MAP), heart rate (HR), CI, CVP, pulmonary arterial wedge pressure (PAWP), systemic vascular resistance index (SVRI). In contrast, oxygen metabolism indexes included SaO 2 , PaO 2 , PH, mixed venous oxygen saturation (SvO 2 ), hemoglobin (Hb), and arterial lactate concentrations obtained from blood sample analysis at every predened study point. The primary outcome was the positive response rate to FC at different time points during the study period. The secondary outcomes were the time course of hemodynamic variables before and after FC. We also explored the effect of uid type (gelatin and NS) on the FR In this scenario.


Introduction
Evaluation of uid responsiveness (FR) is an essential maneuver in the management of uid therapy in critically ill patients 1 . When FR was evaluated, only 40-50% of all these patients, including septic patients, responded 2 . On the other hand, uid overload is harmful and associated with poor prognosis [3][4][5] .
Therefore, some techniques for evaluating FR, such as uid challenge (FC), are recommended to be rst used when hypovolemia or preload dependency is suspected. [6][7][8] . FC refers to giving a certain amount of intravenous uid quickly to assess the response of cardiac output (CO) to uid infusion 9 . Although FC has been considered the gold standard for FR assessment, there is a wide divergence in performing it 7,8 . Among them, timing assessment of response to FC raises lots of concerns.
Most studies only reported a certain assessment time point of response to uid infusion, ranging from the end of FC to 30 min thereafter 7,8 . However, hemodynamics after FC may change over time, while the persistence of FR after FC has been poorly described 8, 10 . On the other hand, the degree and durations in intravascular distribution caused by different types of uids (e.g., crystalloid or colloid) may also vary 11,12 . Additionally, some clinical factors such as blood volume status, cardiac function, and capillary leak severity of septic shock patients might further complicate the FC procedure 13,14 .
In some previous studies that focused on post-cardiac surgery patients receiving FC by 6% hydroxyethyl starch, nearly 74% (56-90%) of included patients were deemed uid responders when evaluated 10 min after FC [15][16][17][18][19] . This proportion was much higher than that in most other studies (54%) assessing FR immediately after infusion 8 . Therefore, we could speculate that some patients who had shown immediate negative response at the end of infusion might become responders at 10 min after FC. In the FCREV study 20 , Roger and colleagues also observed this phenomenon; among the 67 non-responders at the end of FC, 4 became responders 20 min later. Moreover, Aya et al. enrolled 26 postoperative patients in their FC study (250 ml crystalloid infusion over 5 min) and found the effect of uid infusion dissipated in 10 min in these patients 21 . However, a recent meta-analysis focusing on FC showed 13% (11/85) of the included studies evaluated FR at a time point of ≥10 min after the uid bolus 8 . As such, the proportion of responders may be different when evaluating FR at different assessment time points. Therefore, we sought to perform a prospective cohort study to determine the optimal assessment time after FC in patients with septic shock. Additionally, we further compared the effect of different uids on FR, such as gelatin or normal saline (NS), by investigating the hemodynamics for 120 min after FC. We hypothesized that optimal assessment time might be different when using different types of uids.

Study protocol
This prospective observational study was approved by the Ethical Committee of Peking Union Medical College Hospital and was registered at ClinicalTrial.gov (NCT01941472). Written informed consent was obtained from all patients or their relatives. We performed this study in our 15-bed medical intensive care unit (ICU) between September 2015 and November 2018. A owchart of the study design is shown in Additional le 1.

Patients
All adult patients were eligible if they were diagnosed with septic shock according to international criteria 22 and required an FC during their stay in ICU. We excluded patients known with severe cardiac dysfunction (i.e., acute pulmonary edema, acute coronary syndrome,and cardiogenic shock), pregnancy, and participation in other biomedical studies, requiring blood transfusion or imminent death within 24 hours. Also, we excluded patients who required aggressive uid therapy ( uid infused >200 mL/h), dose changes in sedatives, inotropic or vasoactive agents, or adjustments to ventilator parameters during the study period.

Hemodynamic monitoring
We monitored arterial blood pressure from an arterial line (Becton Dickinson infusion therapy systems Linc., Utah, USA) placed in a radial artery or dorsalis pedis artery; and measured CVP with a central venous catheter (CV-15854; Arrow International, Reading, PA) inserted into the internal jugular vein in all patients. CVP and blood pressure were measured with a transducer zeroed at the level of the midaxillary line of the thorax. The CI was calculated by the continuous thermodilution technique equipped with a PAC (Swan-Ganz CCOmbo CCO/SvO 2 , Edwards Lifesciences, Irvine, CA, USA). We connected all the above catheters to pressure transducers and the IntelliVue Patient Monitor MP70 (Philips Medical System, Boeblingen, Germany).

Fluid challenge
According to our protocol, reasons for FC included hypotension (SBP ≤90 mmHg or SBP decrease ≥40 mmHg in patients with hypertension or MAP ≤65 mmHg), presence of tissue hypoperfusion (including, but not limited to, oliguria, skin mottling, cool peripheries, altered mental status, hyperlactatemia, and increased requirement for catecholamine). For the FC administration, 500 ml of 4% gelatin (Gelofusine; B. Braun Medical (Suzhou) Company Limited, Suzhou, China) or NS were intravenous infusions over 5-10 min using a bag pressurized to 300 mmHg. Positive FR was de ned as an increase in CI >10% after FC 9 . The uid infused, starting and termination of FC were decided by clinicians. The patients were followed during FC and for 120 min after FC. During the observational period, the maintenance infusions were limited to a maximum of 100 mL/h, without any changes made to agents, ventilatory settings, and other therapeutic interventions.

Parameters and outcomes
Once enrolled in our study, demographics, uid types, Acute Physiology and Chronic Health Evaluation II score, underlying diseases, and clinical data concerning therapies (mechanical ventilation, renal replacement therapy, sedatives, catecholamine used) were collected for included patients. We recorded a complete set of hemodynamic and oxygen metabolism variables for analysis at baseline (T b ), immediately (T 0 ), and at 10 (T 1 ), 30 (T 2 ), 45 (T 3 ), 60 (T 4 ), 90 (T 5 ) and 120 (T 6 ) min after FC. Hemodynamic variables included mean arterial pressure (MAP), heart rate (HR), CI, CVP, pulmonary arterial wedge pressure (PAWP), systemic vascular resistance index (SVRI). In contrast, oxygen metabolism indexes included SaO 2 , PaO 2 , PH, mixed venous oxygen saturation (SvO 2 ), hemoglobin (Hb), and arterial lactate concentrations obtained from blood sample analysis at every prede ned study point.
The primary outcome was the positive response rate to FC at different time points during the study period. The secondary outcomes were the time course of hemodynamic variables before and after FC. We also explored the effect of uid type (gelatin and NS) on the FR In this scenario.

Statistical analysis
Categorical variables were expressed as numbers (%), whereas continuous variables were expressed by the means±standard deviation (SD) or by the medians along with 25-75% interquartile ranges (IQR), as appropriate. Before FC, patient characteristics between gelatin and NS groups were compared using Student's T or Mann-Whitney U for continuous data and the chi-square test or Fisher's exact tests for categorical data, according to their distribution. After FC, a two-way repeated measure analysis of variance was used to compare the hemodynamic and metabolism variables among the patients grouped by uid types or FR status at all the prede ned time points. MedCalc statistical software version 15.6.1 for Windows and GraphPad Prism 7 were used in the present study. A 2-sided P value less than 0.05 was considered statistically signi cant.

Patient characteristics
Sixty-three patients were nally recruited (Additional le 2), with 43 patients in the gelatin group and 20 in the NS group. Patient baseline characteristics and clinical data were comparable between the groups (Table 1). All patients had a mean age of 63 years with a mean APACHE II score of 25 at enrolment.
Hypotension was the most common cause of FC, following by increased catecholamine requirement and hyperlactatemia. Twenty-nine patients died during their ICU stay.

Fluid responsiveness assessment
Forty-ve patients were responders,with 31 in the gelatin group and 14 in the NS group. As to the 18 persistent non-responders (NR), 12 were in the gelatin group and 6 in the NS group. Among the responders, 31 responded immediately at T 0 , while 14 became responders at or after T 1 . Thus, we de ned these patients as early responders (ER) and later responders (LR), respectively. The baseline clinical data of NR, ER, and LR are summarized in Additional le 3.

Hemodynamic assessment
Hemodynamic variables at each prede ned time point were summarized according to uid types (gelatin and NS), or FR status (NR, ER, and LR) are shown in Additional le 4 and Table 2, respectively. Overall, CI increased signi cantly in both gelatin and NS groups at T 0 , with different trends over time. In the gelatin group, CI increased to a maximum at T 2 and maintained an increase in CI >10% before T 4 . In the NS group, CI increased to a maximum at T 1 and decreased to baseline at T 3 ( Fig. 2A). Subsequently, we further explored CI changes over time in NR, ER, and LR subgroup patients. In the gelatin group, ER and LR became responders at T 0 and T 1 , respectively, and then achieved CI max at T 2 and maintained positive FR until T 6 (Fig. 2B). As to the NS group, ER and LR responded at T 0 and T 1 , respectively, and then presented CI max at T 1 . Finally, ER and LR changed their FR status from R to NR at T 3 and T 2 . CI did not signi cantly increase over time in NR of both groups after FC (Fig. 2C). The time courses of other hemodynamic variables before and after FC are shown in Additional le 5.  T 6 = 120 min after FC.

Discussion
The main ndings of our study are, rst, the proportion of NR, ER, and LR were comparable between gelatin and NS groups; second, FC induced the most frequency of CI max at T 2 and T 1 in gelatin and NS groups, respectively, and the positive FR status sustained until T 4 in gelatin group, while until T 1 in NS group; third, 14 patients identi ed as NR at T 0 became LR at T 1 or later, and both ER and LR achieved CI max at T 2 after FC in gelatin group, whereas at T 1 in NS group.
Our results showed gelatin exhibited a longer duration of positive response status than NS after FC. This might be due to the difference in hemodynamic effects between crystalloid and colloid. That is, colloid can maintain a longer-term hemodynamic effect when compared with crystalloid in equal scenario 11,23 . In a recent randomized trial, Gondos and colleagues examined the kinetics of volume loading with crystalloid and colloid infusions in 200 critically ill patients, and they found that 6% hydroxyethyl starch still produced a change in CI (23%) at 120 min after infusion, while this effect dissipated in lactated Ringer's solution 11 .
On the other hand, positive response status over time to FC may also be affected by patient populations and details in performing FC (i.e., a CO monitoring technique, duration or volume of infusion) 20 Interestingly, 22% (14/63) of patients showed an initial negative response immediately after FC subsequently converted to LR. We called these patients "late responders to FC." Such a phenomenon was observed in septic shock patients 20 or post-cardiac surgery patients [15][16][17][18][19] in some previously published studies. These patients are likely to be overlooked when FR is only evaluated immediately at the end of the infusion. Thus, this nding supports our hypothesis that timing assessment of FR could affect the proportion of patients responding to FC. However, our conclusion is in contrast with a recently published metaanalysis that focused on FC 8 . In this study, the authors grouped the 86 included studies into three categories based on assessment time (immediately, between 1 and 10 min, or >10 min after FC) and found FR timing assessment did not affect the proportion of responders.  11 . Second, the volume of uid used was 500 ml, the current "mainstream volume" selection for FC 26 . However, whether our results could also be applied for other xed uid volumes (i.e., 250 ml) or uid volume adjusted for body-weight remains unclear. This would require further investigation. Third, the process of FC was completed in about 6 min, which is faster than most other FC studies 8, 10 . Theoretically, the shorter infusion duration, the larger proportion volume of uid will remain in the intravascular compartment at the end of infusion, and a higher positive FR rate may be obtained. This was also con rmed by the meta-analysis result, which demonstrated that the proportion response to an FC given in ≥30 min was lower than that in <15 min (P=0.045) 8 . Thus, this might indirectly add evidence for the reliability of negative FR in LR patients immediately after FC in the current study. Finally, we explored the time course of FR in 120 min after uid infusion, a relatively longer period compared to the previous studies 10,20,21 . Apart from the purpose of a complete recording of hemodynamic effects on FR, the main consideration for this is the tolerability of volume-limited (infusion <100 mL/h) during 120 min in the enrolled patients. Overall, no adverse events were observed during the study period, which may be related to the initial uid resuscitation performed in these patients before enrollment.
Our study had several limitations. The rst and main limitation is the incapability to fully explain our results from the pathophysiology mechanism due to the pure observation nature of the current study, especially for LR patients. Nevertheless, our data may, at the very least, encourage clinicians to reevaluate their practice in deciding timing assessment of response to FC in septic shock patients. Second, FR was evaluated at only 7 prede ned time points referred to most previous studies 8, 10,11,27 . However, our results suggest that more assessment time points may be essential in the early stage after FC, especially in crystalloid group patients with relatively short periods of hemodynamic effect. Finally, only septic shock patients having received initial uid resuscitation were included according to the current protocol. However, the fully initial uid resuscitation has not been strictly de ned and lacked individuation. A previous study reported FC (5 ml/kg crystalloid solution infusion over 15 min) induced a signi cant increase in CO and sustained for 120 min in severe hypovolemic sepsis patients without initial uid resuscitation. Whether our results could apply to such a patient population is unclear.

Conclusion
Signi cant differences were found in time courses of hemodynamic effects in septic shock patients receiving 500 ml gelatin or NS for FC. Therefore, the timing assessment of response to FR differs between the two types of uids. When NS is used, FR should be performed from the end of FC to 10 min that after, while it is better to extend assessment time to 30 min after FC in gelatin, especially for patients who show negative responses immediately after infusion.

Declarations
Author contributions: Dr. Huang contributed to the data collection, analysis, and drafting of the article. Dr. Liu and Dr. Xu contributed to data collection, literature search, and writing of the manuscript. Dr. Du was responsible for the work integrity as a whole, from inception to publication of the article.