The Monitoring Paradox: Frontal Electroencephalography Reveals Profound Cortical Depression Despite Cardiovascular Stability in Dogs Undergoing Tibial Plateau Leveling Osteotomy

1. Introduction

General anesthesia is a drug-induced, reversible state defined by unconsciousness, amnesia, immobility, and reduced autonomic responses to noxious stimulation [1,2,3]. Although anesthetic agents act broadly throughout the central nervous system, the cerebral cortex is far more sensitive to anesthetic-induced depression than subcortical structures such as the brainstem and thalamus [4,5,6]. Clinical and experimental evidence shows that anesthetics primarily disrupt cortical communication required for conscious awareness while still allowing nociceptive and sensory signals to reach the brain [2,6,7,8]. Consequently, modern anesthetic practice treats hypnosis and antinociception as distinct therapeutic targets, each requiring individualized drug titration to modulate cortical and subcortical networks during surgery [2,3,4,5].

Loss of consciousness arises largely from fragmentation of cortical connectivity [5,6,7]. Contemporary systems neuroscience further clarifies that anesthesia operates through two interrelated dimensions: degradation of the content of consciousness through top-down frontoparietal networks and suppression of the level of consciousness through bottom-up arousal pathways [6,8,9]. Despite this dual framework, veterinary monitoring continues to rely heavily on autonomic reflexes mediated by the brainstem, including heart rate, respiratory rate, and arterial pressure [1,10,11]. These variables serve as indirect indicators of antinociception rather than direct measures of cortical hypnosis [1,11]. The 2025 ACVAA monitoring guidelines emphasize physical signs such as muscle tone, eye position, reflex responses, and end tidal inhalant concentration [12]. These parameters primarily reflect spinal and brainstem physiology [4,5,10]. Because subcortical structures are evolutionarily resistant to anesthetic depression, they may remain stable even when cortical activity is profoundly suppressed [4,5,6,7,13]. This creates a monitoring paradox in which subcortical homeostasis appears preserved, while the cortex, the primary pharmacologic target, undergoes deep, and often unrecognized, depression [5,6,11].

Human anesthesia has addressed this gap by incorporating processed electroencephalography (EEG) and Parasympathetic Tone Activity (PTA) to directly quantify cortical and autonomic dynamics [1,2,5,13,14,15]. EEG derived indices such as the Bispectral Index (BIS), entropy, and the Patient State Index (PSI), together with raw EEG and spectrograms, provide real time evaluations of cortical state and help prevent awareness and excessive anesthetic exposure [1,2,5,6,16]. PTA complements these measures by quantifying beat to beat autonomic responsiveness to nociception [14,15,17]. Collectively, these modalities offer a multimodal characterization of brain state that traditional hemodynamic monitoring cannot capture [5,7,13].

Distinct EEG signatures produced by different anesthetic classes further illustrate the mechanisms of unconsciousness [3,5,7]. Volatile agents, propofol, and opioids generate characteristic patterns that reflect their effects on cortical connectivity, whereas dissociatives such as ketamine may increase cortical activity despite producing unconsciousness [3,5,6,7]. Profound cortical suppression, including burst suppression, is associated with delayed emergence and postoperative cognitive dysfunction in humans [11,18,19,20,21]. These insights highlight the value of direct cortical monitoring, especially in multimodal protocols where synergistic drug interactions can produce deep cortical depression despite stable hemodynamic variables [5,6,7,22].

In canine orthopedic procedures such as Tibial Plateau Leveling Osteotomy (TPLO), multimodal protocols commonly combine alpha 2 agonists, dissociatives, potent opioids, and regional anesthesia. Systemic opioids and sedatives substantially reduce the inhalant concentration required for immobility, increasing the risk of a Triple-Low state characterized by hypotension, low inhalant concentration, and profound cortical suppression [3,5,11,23,24,25,26]. In humans, Triple-Low states are associated with delayed extubation, emergence delirium, and impaired recovery trajectories [11,18,19,20,27]. Clinically, dogs undergoing these procedures often experience prolonged or difficult recoveries that require antagonism of alpha 2 agonists or opioids. Combined with the known sensitivity of cortical networks to anesthetic induced disconnection, these observations suggest that important aspects of anesthetic depth may go unrecognized when relying solely on hemodynamic variables [1,5,6,7].

This study was designed to determine whether frontal EEG and PTA, recorded in parallel with standard cardiorespiratory monitoring but not used to guide anesthetic management, could reveal unrecognized gaps in depth of anesthesia assessment that are not apparent from conventional variables. To evaluate dogs under different analgesic conditions, three commonly used techniques involving systemic opioids or regional approaches were compared. We hypothesized that EEG and PTA would identify low cortical activity, burst suppression, and depressed brain state patterns beyond what cardiorespiratory variables detect. We further hypothesized that dogs receiving systemic opioid based analgesia would exhibit more suppressed intraoperative cortical states and more challenging recoveries than dogs receiving regional techniques based on local anesthetics.

2. Materials and Methods

2.1. Animals and Study Design

Twenty-one client-owned dogs presented to the Purdue University Veterinary Hospital for unilateral TPLO were enrolled in this prospective randomized clinical trial. Written informed consent was obtained from all owners. The study protocol was approved by the Purdue University Institutional Animal Care and Use Committee (PACUC #0724002513PA277) and conducted in accordance with institutional ethical guidelines and regulations. All dogs were classified as American Society of Anesthesiologists Physical Status II based on clinical history, physical examination, complete blood count, and serum biochemistry. Dogs were excluded if they had pre-existing neurological disease, cardiac arrhythmias, skin allergy above the surgical site, or had received analgesic or sedative medications within 24 hours of surgery.

2.2. Anesthetic and Analgesic Protocols

Dogs were randomly assigned to one of three analgesic treatment groups (n = 7 per group) using a computer-generated randomization sequence prior to the start of the study. The three treatment groups consisted of an intravenous fentanyl constant rate infusion (FC), a lumbosacral epidural morphine injection (ME), and an ultrasound guided regional block consisting of sciatic and saphenous nerve blocks (RB). The FC group received an intravenous fentanyl (Fentanyl 0.05 mg/mL Injectable, Hospira, Lake Forest, IL, USA) loading dose of 5 µg/kg followed by a constant rate infusion of 5 µg/kg/h. The ME group received a lumbosacral epidural injection of preservative free morphine (Morphine Sulfate 1 mg/mL, Hospira, Lake Forest, IL, USA) at 0.1 mg/kg, diluted 1:1 with 0.9% sodium chloride (Hospira, Lake Forest, IL, USA) to a total volume of 0.2 mL/kg, administered using a 22 gauge spinal needle (BD, Franklin Lakes, NJ, USA or RapID™, Irving, TX, USA). The RB group received ultrasound guided sciatic and saphenous nerve blocks using 0.5% bupivacaine (Hospira, Lake Forest, IL, USA) at 0.1 mL/kg per site, performed with a Mindray TE7 Max Advance ultrasound unit (Mahwah, NJ, USA) and a 21G × 4″ echogenic needle (Medline^®, Northfield, IL, USA). All epidural injections and ultrasound guided regional nerve blocks were performed by a single clinician to maintain procedural uniformity across cases. All analgesic interventions (fentanyl bolus/CRI initiation, epidural injection, and both regional nerve blocks) were completed within 8 minutes of anesthetic induction.

All dogs received a standardized anesthetic and perioperative regimen. Premedication consisted of intramuscular dexmedetomidine (5 µg/kg, Dexdomitor 0.5 mg/ml, Zoetis, Kalamazoo, MI, USA) and hydromorphone (0.1 mg/kg, Hydromorphone 2 mg/ml, Hikma, Berkeley Heights, NJ, USA). After approximately fifteen to twenty minutes, a cephalic vein was catheterized (IV catheter, Terumo™, Somerset, NJ, USA), and anesthesia was induced with propofol and ketamine. The induction was performed by administering 1 mg/kg propofol (Propoflo 10 mg/ml, Zoetis, Kalamazoo, MI, USA) followed immediately by ketamine (2 mg/kg, Ketamine 100 mg/ml, Dechra, Overland Park, KS, USA), then titrating the additional 1 mg/kg propofol to effect to allow endotracheal intubation. Balanced electrolyte fluids (pHyLyte, Dechra, Overland Park, KS, USA) were administered at 5 mL/kg/h throughout anesthesia.

Anesthesia was maintained with isoflurane (Fluriso, Vet One, Austin, TX, USA) in 100% oxygen using an anesthesia machine and ventilator (Mindray WATO Ex 35, Mahwah, NJ, USA). All dogs were mechanically ventilated in pressure-controlled mode to maintain end tidal carbon dioxide (Et-CO₂) between 35 and 45 mmHg. The end tidal isoflurane (Et-Iso) concentration was initially set between 1.0 and 1.3% and subsequently adjusted based on conventional clinical indicators of anesthetic depth, including eye-globe position, corneal and palpebral reflexes, jaw tone, and hemodynamic stability. EEG and PTA were recorded throughout anesthesia for observational purposes only and did not influence anesthetic management. At the end of surgery, isoflurane was discontinued, and the time to extubation was recorded. Extubation was defined as the return of the swallowing reflex followed by removal of the endotracheal tube.

2.3. Cardiorespiratory and Electroencephalographic Monitoring

A multiparameter monitor (Mindray Multiparameter Monitor, Mahwah, NJ, USA) was used for all physiologic monitoring. Heart rate and cardiac rhythm were assessed using a lead II ECG. Pulse rate (HR) was obtained from pulse oximetry, and direct arterial blood pressure, including systolic blood pressure (SBP), mean blood pressure (MBP), and diastolic blood pressure (DBP), was measured via an arterial catheter (Terumo™, Somerset, NJ, USA) placed in the dorsal pedal artery. Et-CO2 and Et-Iso concentrations were measured using a side-stream capnograph with anesthetic agent monitoring capability (Mindray Side-stream Capnograph, Mahwah, NJ, USA). Body temperature was monitored with an esophageal probe during surgery and transitioned to a rectal probe after extubation. All instrumentation was completed within 10 minutes after induction, and physiologic variables were recorded every 5 minutes.

Surrogate indicators of anesthetic depth were also assessed, including eye globe position, corneal and palpebral reflexes, jaw tone, and front-limb muscle relaxation. Frontal EEG was recorded using a modified SedLine^® system adapted for canine skull morphology [28,29]. Adult SedLine^® adhesive electrodes were converted to subdermal needle electrodes (Ambu, Columbia, MD, USA) to improve contact and avoid clipping. Before each procedure, electrodes were tested and verified by the SedLine Root^® monitor, which requires impedance values between 0.0 and 65.1 kΩ. Electrodes were positioned analogously to the human 10–20 system following the established canine montage: R1 at Fp2, R2 between F4–F8, L1 at Fp1, and L2 between F3–F7. The ground (CB) electrode was placed along the mid-sagittal line, and the reference (CT) electrode was placed cranially along the same line. All EEG-derived indices, including the Patient State Index (PSI), Suppression Ratio (SR), 95% Left and Right Spectral Edge Frequency (SEF-L and SEF-R), and Density Spectral Array (DSA) parameters, were continuously recorded. High-resolution EEG data were exported as CSV files at 2-second intervals and downloaded from the Sedline monitor.

PTA monitoring was performed using the MDoloris system (Loos, France), which analyzes beat-to-beat R R interval variability. The PTA device uses a three-lead ECG configuration with alligator clips, integrated into the existing lead II setup. A green icon confirmed adequate signal acquisition. Only the mean PTA (mPTA) was used for analysis. The mPTA is calculated from the most recent 176 R R intervals, representing a 64 second moving window. mPTA values were recorded every 5 minutes. On the PTA scale (0–100], values between 50 and 70 indicate optimal analgesia–nociception balance, whereas values below 50 reflect sympathetic predominance.

2.4. Recovery Assessment and Postoperative Analgesic Protocol

Recovery evaluation began at the termination of isoflurane and continued through emergence and extubation. Recovery quality was assessed using a simple descriptive scale: 1 (excellent), 2 (good), 3 (acceptable), 4 (poor), and 5 (very poor). Emergence was classified as “smooth” for scores of 1, 2, or 3, and “rough” for scores of 4 or 5. After extubation, postoperative pain was assessed at 1, 2, 4, 8, 12 hours, and before discharge (<12 hours) using the Glasgow Composite Measure Pain Scale–Short Form (CMPS-SF) [30]. Rescue analgesia consisted of hydromorphone (0.05 mg/kg IV; Hydromorphone 2 mg/ml, Hikma, Berkeley Heights, NJ, USA) and was administered if the CMPS-SF score [23] exceeded 6/24 (or 5/20 for non-ambulatory dogs). Both the frequency of rescue analgesia and total postoperative opioid consumption were recorded.

2.5. Data Processing and Statistical Analysis

2.5.1. Data Integration and Time-Point Standardization

To account for variation in surgical duration, all physiologic, hemodynamic, EEG, and nociception-related data were aligned to a standardized procedural timeline. Processed EEG indices were sampled every 2 seconds, while hemodynamic variables and mPTA were recorded at 5-minute intervals. The anesthetic and surgical period was divided into twelve clinically defined intervals that reflect consistent procedural milestones across all dogs. Time point 1 (T1) corresponded to premedication, followed by Time point 2 (T2) for induction and intubation. Draping and surgical readiness were completed at Time point 3 (T3), and arthroscopy was performed at Time point 4 (T4). The first skin incision was designated Time point 5 (T5), while Time point 6 (T6) encompassed muscle and fascia dissection, as well as periosteal elevation. Bone drilling, bone sawing, and osteotomy were grouped within Time point 7 (T7), and pin or implant placement occurred during Time point 8 (T8). Suturing and closure of the muscle and skin were represented at Time point 9 (T9), with the end of the procedure captured at Time point 10 (T10). Termination of inhalant anesthesia occurred at Time point 11 (T11), and extubation occurred at Time point 12 (T12). Values within each interval were averaged to generate a single representative value for statistical modeling.

2.5.2. Clinical Definitions and Descriptive Analysis of Low States

In contrast to the interval-based averaged values described above, Double-Low and Triple-Low states were identified from the unaveraged, continuously sampled physiologic and anesthetic data (raw data), evaluated across three domains: cortical activity, arterial pressure, and inhalant concentration. The definitions used in this study were adapted from the human perioperative anesthesia literature [9,10,11,12,18], in which concurrent low cortical activity, hypotension, and low volatile anesthetic concentrations have been associated with adverse postoperative outcomes. In our adaptation, low cortical activity was defined by a PSI ≤ 25, with SR ≥ 20% indicating a Low Brain State; low arterial pressure was defined as mean blood pressure (MBP) ≤ 60 mmHg; and low inhalant concentration was defined as end tidal isoflurane ≤ 1.02%, corresponding to approximately 0.8 MAC in dogs based on established canine MAC values of 1.28–1.30% [23,24,25,26]. Each raw data point within each time interval was reviewed, and concurrent alignment of any two criteria was classified as a Double-Low, whereas alignment of all three criteria was classified as a Triple Low. These classifications were applied uniformly across all dogs and all recorded intervals.

2.5.3. Statistical Analysis

The hemodynamic, anesthetic, EEG, and PTA variables were analyzed across the twelve clinical time points using linear mixed effects models. These included HR, SBP, MBP, DBP, Et-Iso, PSI, SR, SEF-L, SEF-R, and mPTA. Analgesic treatment group, time point, and their interaction were initially included as fixed effects, with dog modeled as a random intercept to account for repeated measurements. The treatment by time interaction was examined to determine whether the three analgesic techniques differed in their hemodynamic, EEG, and PTA profiles over time. When the interaction was significant, comparisons were made within groups or time points. When the interaction was not significant, the interaction term was not interpreted, and main effect trends were summarized.

Contrasts were specified to compare each intra-anesthetic time point (T2 through T11) with the premedication (T1) and extubation (T12) reference points, when data were available. Additional successive time point contrasts were used to evaluate changes across key procedural transitions, particularly from surgical readiness to surgical stimulation and from the end of surgery into recovery. Holm adjustment was applied within each outcome to control the family-wise error rate for these planned temporal contrasts. Contrast estimates with 95% confidence intervals (CI) were reported to describe the magnitude and direction of observed effects. Post hoc comparisons among analgesic treatment groups were performed only after significant model effects were detected, with Tukey adjustment for pairwise differences.

Recovery quality (smooth vs. rough) and rescue analgesia use were evaluated using Fisher’s exact test. Double Low and Triple Low events were recorded as binary variables at each clinical time point (T1–T12), and mixed-effects logistic regression was used to compare event occurrence across analgesic techniques and over time. Outcomes with very few events or insufficient between dog variability were summarized descriptively. Statistical significance was set at p < 0.05. All analyses were performed in R (Version 4.6.0; R Core Team, 2026) and RStudio (Version 2026.07; Posit Team, 2026).

3. Results

3.1. Patient Demographics and Procedural Durations

A total of 21 dogs were enrolled, with no significant differences in demographics (age, weight, sex) observed across the three treatment groups (Table 1). Surgical duration did not differ significantly among groups (mean range: 108–124 min; p = 0.612), nor did total anesthesia time (mean range: 231–274 min; p = 0.160).

3.2. Hemodynamic and Autonomic Trends Across TPLO Time Points

All physiologic and hemodynamic variables demonstrated significant temporal fluctuations throughout anesthesia and surgery (all p < 0.001). The trend could be observed in Figure 1 and Figure 2. However, the main effect of the treatment group was not significant for any parameter (all p > 0.05). HR, DBP, and mPTA did not demonstrate significant treatment-by-time interactions (all p > 0.05) (Figure 1, Figure 2 and Figure 3), and their overall temporal patterns were evaluated irrespective of treatment group. For HR (Figure 1), significant temporal changes were detected between extubation (T12) and the termination of isoflurane anesthesia (T11) (p < 0.001), as well as between T12 and the premedication (T1) (p < 0.001). No significant differences in Et-Iso concentration were observed among treatment groups (Figure 1).

SBP and MBP (Figure 2) demonstrated significant treatment-by-time interactions (p = 0.004 and p = 0.026, respectively). While these pressures remained within clinically acceptable ranges for all dogs, evaluating the data at individual time points revealed distinct group differences. At the end of surgery (T10), MBP was, on average, 15.0 mmHg (95% CI: -30.1, 0.16; p = 0.053) lower in the FC group than in the RB group (Figure 2). SBP was significantly higher in the RB group than in the FC group during joint closure (T9 mean difference: 27.8 mmHg; 95% CI: 3.26, 52.3; p = 0.024) and at T10 (mean difference: 31.0 mmHg; 95% CI: 5.74, 56.2; p = 0.013) (Figure 2).

3.3. Electroencephalographic Trends over the TPLO Procedure

Cortical EEG indices (Figure 3 and Figure 4), specifically the PSI and SR (Figure 3), demonstrated significant treatment-by-time interactions (p = 0.003 and p = 0.050, respectively). Within the FC group, temporal contrasts revealed a significant decline in PSI from T1 to initial surgical readiness (T3) (mean difference: -53.8; 95% CI: -76.2, -31.4; p < 0.001). In contrast, this decline was smaller in the RB group at the same transition (mean difference: -23.2; 95% CI: -48.0, 1.63; p = 0.334). Within the FC group, significant temporal differences in SR were also observed from T4 to T10 relative to T1, with a significant increase (p = 0.004) from the end of surgery (T10) to the termination of isoflurane anesthesia (T11) (Figure 3). No significant within-group temporal variations in SR were observed for the RB or ME groups. Bilateral 95% Spectral Edge Frequencies (SEF-L and SEF-R) (Figure 4) and Et-Iso concentrations did not demonstrate significant treatment-by-time interactions.

3.4. Incidence of Double and Triple-Low States

A total of 71 Double-Low events (Table 2) were identified, consisting entirely of cortical suppression (Low Brain State) paired with low inhalant requirements. Mixed-effects logistic regression indicated that the incidence of Double-Low states (Figure 5) varied significantly by treatment group (p = 0.025). The FC group demonstrated a significantly higher incidence of Double-Low states compared with the ME group (Odds ratio: 7.26; 95% CI: 1.21, 43.5; p = 0.024) (Figure 5). No significant differences in Double-Low incidence were detected between the RB and FC groups (p = 0.153) or between the RB and ME groups (p = 0.339).

3.5. Emergence and Immediate Recovery

During the immediate recovery phase, emergence times were comparable among the protocols, with median durations of 19.0 minutes for the FC group, 19.0 minutes for the ME group, and 16.0 minutes for the RB group. At the end of surgery (T10), the FC group showed the greatest cortical suppression, with higher SR and lower PSI than the ME group; however, neither difference was statistically significant (SR mean difference: 32.3; 95% CI: −6.23, 70.7; p = 0.117; PSI mean difference: −30.1; 95% CI: −64.1, 3.94; p = 0.093). By the termination of inhalant anesthesia (T11), EEG profiles were similar across treatments, with between-group differences in mean SR and PSI no greater than 4.6 and 7.9, respectively. Upon extubation, 5 of 7 dogs in the RB group recovered smoothly. Rough, delirious recoveries were observed in 4/7 dogs in the ME group, 3/7 in the FC group, and 2/7 in the RB group (p = 0.854). Rescue sedation or analgesia was administered to 4 dogs in the ME group, 3 in the FC group, and 2 in the RB group during the immediate postoperative period.

3.6. Postoperative Pain and 24-Hour Rescue Analgesia

In the 24-hour postoperative period, median peak Glasgow Composite Measure Pain Scale–Short Form scores were 3 in the FC group, 2 in the ME group, and 3 in the RB group. Postoperative rescue analgesia interventions were required for 3 of 7 dogs in the FC group (3 total doses) and 2 of 7 dogs in the ME group (4 total doses), while no dogs in the RB group reached the clinical threshold for rescue analgesia (p = 0.292).

3.7. Perioperative Brain State Progression Through Electroencephalography

To visually illustrate these neurophysiological transitions, Figure 6, Figure 7 and Figure 8 provide representative SedLine monitor screenshots capturing distinct brain states in an FC-treated dog. This specific patient (Dog ID 5) was selected as a representative case because its documented clinical emergence directly mirrored the profound neurophysiological shifts captured by EEG monitoring. Clinically, this dog experienced a rough recovery characterized by vocalizations, necessitating rescue sedation with hydromorphone and dexmedetomidine. The provided image sequence tracks the progression from an initial ketamine-dominant phase following propofol-ketamine induction, through deep intraoperative cortical depression characterized by high burst suppression ratios during isoflurane and fentanyl maintenance, and ultimately to the abrupt cortical awakening that preceded this rough early recovery.

4. Discussion

The results of this study demonstrate that frontal EEG and PTA monitoring provide clinically meaningful insight into cortical state and autonomic balance during canine TPLO procedures. These findings also reveal a substantial gap in routine anesthetic assessment: dogs frequently exhibited marked cortical suppression even when hemodynamic variables remained within accepted clinical ranges.

The monitoring paradox arose because autonomic variables remained stable even when cortical activity showed substantial suppression. The groups displayed similar HR, Et-Iso, SBP, MBP, DBP, and PTA values, indicating preserved cardiovascular homeostasis supported by brainstem autonomic centers [1,3,7,14,15,16,17]. In contrast, the processed EEG revealed deep cortical depression, with the FC group demonstrating the greatest susceptibility. This pattern became most apparent at the onset of surgical stimulation. At T3, PSI decreased from T1 by 53.8 in FC dogs, 51.9 in ME dogs, and 23.2 in RB dogs. SR values increased by 31.8%, 22.8%, and 1.6% in the FC, ME, and RB groups, respectively. These findings show that the transition from pre-incisional conditions to early surgical stimulation produced measurable shifts in cortical state across all groups, and that the FC protocol resulted in the deepest and most consistent cortical depression despite similar autonomic profiles.

General anesthetics depress subcortical arousal systems while disrupting cortical networks responsible for the content of consciousness, particularly frontoparietal circuits involved in large scale integrative processing [3,6,7]. They also impair reciprocal communication between the cortex and thalamus, particularly via matrix cell pathways that coordinate widespread cortical information flow [6,7,20]. Mathematical and computational models show that anesthetics reduce ascending arousal from the brainstem and weaken synaptic connectivity within both cortex and thalamus [3,6,8]. These combined effects produce the EEG slowing and burst suppression patterns observed in this study [5,7,21,22,28,29]. Because anesthetics preferentially disrupt horizontal cortical communication while leaving ascending nociceptive pathways relatively intact [3,6,7], surgical stimuli can still activate subcortical networks. This activation produces catecholamine mediated sympathetic responses that maintain autonomic vital signs even when cortical networks are fragmented [6,7].

PTA monitoring provided additional clarity by continuously tracking heart rate variability and demonstrating that autonomic balance was preserved during surgical stimulation. While EEG recordings showed profound cortical suppression, PTA fluctuations reflected nociceptive input without revealing differences among treatment groups. The preservation of autonomic balance, despite marked cortical depression, underscores the functional separation between brainstem-driven physiological responses and higher cortical integrity. These findings support the notion that subcortical nociceptive processing and cardiovascular homeostasis persist independently of cortical state in this clinical model.

The continuous EEG progression of individual dogs further illustrates this dissociation. A representative FC dog (Dog ID 5) demonstrated an active ketamine dominant cortical state following induction and early isoflurane maintenance (Figure 6A), characterized by high frequency, low amplitude waveforms, consistent with previous work [29]. This pronounced cortical activity was not detectable clinically; the dog appeared relaxed and stable despite substantial EMG activity. As surgery progressed and the fentanyl infusion reached full effect (Figure 6B-Figure 7B), the EEG transitioned into profound cortical depression, culminating in a nearly isoelectric state with a PSI value of 1 and a burst suppression ratio approaching 100%.

To systematically evaluate cortical and autonomic dynamics, we applied the Double-Low and Triple-Low concepts adapted from human perioperative literature [9,10,11,12,18,27]. Low cortical activity was defined as PSI ≤ 25; substantial suppression as SR ≥ 20; low inhalant requirement as Et Iso < 1.02; and hypotension as MBP ≤ 60. These thresholds were selected because human studies consistently show that concurrent low cortical activity, low volatile anesthetic requirement, and hypotension identify periods in which cortical integrity, perfusion, and anesthetic depth are simultaneously compromised, a state that is strongly associated with poor recovery outcomes such as postoperative delirium, prolonged hospitalization, and increased mortality [9,10,11,12,18,27]. In the current study, a PSI ≤ 25 falls below the standard surgical range of 25-50. This target range was recently utilized in human clinical trials specifically to guide volatile anesthetic administration and avoid burst suppression [17,31]. Furthermore, a PSI ≤ 25 corresponds to markedly reduced cortical information processing [5,6,7]. SR ≥ 20 identifies periods of substantial suppression, paralleling human definitions of deep cortical depression [5,6,7]. Et-Iso < 1.02% corresponds to approximately 0.8 MAC in dogs [23,24,25,26]. MBP ≤ 60 mmHg reflects a clinically meaningful reduction in perfusion pressure [1,3]. Together, these criteria provide a species-modified, physiologically grounded adaptation of the human Double-Low and Triple-Low constructs and allow structured evaluation of cortical state, autonomic stability, and anesthetic depth in dogs [3,5,6,7,9,10,11,12,17,18,23,24,25,26,27].

This framework clarifies the severe cortical depression observed in several dogs. In the example, the FC-treated dog (Figure 7A and Figure 77B) showed nearly isoelectric EEG waveforms and high burst suppression while blood pressure, heart rate, and Et-Iso concentrations remained clinically stable. While this dog provides a clear visual example, similar Double-Low states occurred across all treatment groups, indicating that these profound depressions reflected the combined influence of the multimodal anesthetic protocol, individual anesthetic sensitivity, and the specific analgesic technique used. Importantly, this level of cortical suppression was entirely undetectable from outward clinical signs; the dogs appeared appropriately anesthetized and stable despite marked EEG depression. Autonomic variables also showed treatment-specific temporal patterns, but these patterns did not align with the depth or direction of EEG changes. Whereas PSI and SR reflected differences in cortical suppression among protocols, SBP and MBP appeared to reflect cardiovascular compensation and responses to surgical stimulation. Thus, clinically acceptable arterial pressures should not be interpreted as evidence that cortical activity is similarly preserved. These observations demonstrate that multimodal anesthesia can produce deep alterations in brain state that remain invisible when relying solely on hemodynamic stability, and that EEG monitoring is necessary to identify these episodes. This recognition naturally leads to the question of how different analgesic techniques influence cortical activity when layered onto the same anesthetic background.

The multimodal anesthetic protocol allowed us to examine how three analgesic techniques modulated cortical state within the shared background of hydromorphone, dexmedetomidine, ketamine, propofol, and isoflurane. Because FC, ME, and RB intercept nociception at different anatomical levels, their cortical effects diverged substantially [3,5,6,7,22]. FC produced the most pronounced cortical depression, consistent with the expected supraspinal effects of systemic opioid administration. RB and ME produced less marked or less consistent cortical suppression, likely reflecting differences in how peripheral, neuraxial, and systemic analgesic techniques modulate nociceptive input and cortical arousal. These findings support the interpretation that cortical EEG monitoring can reveal treatment related differences in brain state that are not apparent from conventional autonomic variables alone.

Double-Low analysis further highlighted these discrepancies. Across all dogs, 71 Double-Low events were identified. FC dogs exhibited the highest incidence, supporting that systemic opioid administration produces more frequent and consistent cortical depression. RB dogs showed greater inter-individual variability, with two subjects demonstrating notable sensitivity while others remained near T1. ME dogs showed sparse and infrequent Double-Low events [27]. These variations suggest that individual sensitivity to anesthetic combinations may have attenuated expected differences between RB and ME in this small cohort.

Emergence times were comparable across groups, indicating that the depth of intraoperative cortical depression did not prolong extubation. However, EEG profiles at the end of surgery demonstrated substantial differences in cortical state. FC dogs had the most depressed EEG profile at the end of surgery, with very low PSI and high SR. ME dogs appeared less suppressed, whereas RB dogs remained suppressed but with a different profile. By T11, PSI and SR values had become more similar across groups, suggesting that cortical recovery began rapidly after inhalant discontinuation. The key point is not delayed emergence time, but the abrupt transition from markedly suppressed cortical activity to wakefulness in dogs with deeper intraoperative depression, particularly in the FC group [5,6,7,21,22,28,29].

Recovery quality did not differ significantly among groups, but descriptive patterns were clinically meaningful. Rough or delirious recoveries occurred in 3 of 7 FC dogs, 4 of 7 ME dogs, and 2 of 7 RB dogs. Rescue sedation or analgesia was required in 3 FC dogs, 4 ME dogs, and 2 RB dogs. These observations parallel human findings in which intraoperative burst suppression is associated with postoperative delirium and impaired cognitive recovery [5,6,7,18,21,31]. Dogs across all groups transitioned from deep cortical suppression to wakefulness while still under the influence of the multimodal protocol. This rapid shift likely created a neurophysiologic disequilibrium that predisposed these dogs to agitation, disorientation, and impaired sensory integration during emergence [21,22]. This phenomenon is illustrated in Dog ID 5 (Figure 8A and Figure 8B), where a rapid ascent of PSI from deep suppression to high activity corresponded with a rough recovery requiring rescue sedation.

Postoperative pain scores and 24-hour rescue analgesia requirements further contextualize these findings. Peak Glasgow pain scores were low across groups, yet rescue analgesia was required in 3 FC and 2 ME dogs, whereas no RB dogs reached the intervention threshold. The absence of rescue analgesia in RB dogs suggests that more stable cortical activity during surgery may support smoother emergence and more consistent postoperative comfort [30]. Together, these results indicate that the degree of intraoperative cortical suppression, rather than emergence time or autonomic stability, may be a key contributor to the quality of recovery and postoperative neurobehavioral outcomes [5,6,7,18,21,31].

This study has several limitations. Respiratory assessment was limited to Et-CO₂ and Et-Iso because all dogs were mechanically ventilated to optimize PTA acquisition. This prevented the evaluation of spontaneous ventilation and masked the respiratory depressive effects of opioids, which are commonly used as clinical indicators of anesthetic depth [3,22]. Although an a priori power analysis was conducted to ensure adequate power for the primary outcomes, the cohort size of 7 dogs per group remains relatively small. This limited sample size, coupled with inherent inter-individual variability in anesthetic sensitivity, likely contributed to a lack of statistical significance in some between-group comparisons. Nevertheless, this sample size successfully captured robust, highly significant temporal shifts in cortical state across the entire cohort, validating the primary premise that these monitors reliably detect clinically meaningful neurophysiological changes. Blinding of anesthesiologists was not feasible due to practical constraints; however, potential observer bias was minimized as the primary outcome measures (PSI, SR, and PTA) were derived directly from automated, objective algorithms. Additionally, because this study used proprietary algorithms specific to the SedLine and PTA monitors, the absolute values and Double-Low thresholds established here may not be directly comparable to those of other depth-of-anesthesia or nociception monitors. Furthermore, while the numerical thresholds defining the canine Double-Low state were grounded in physiological principles and human literature, they have not yet been prospectively validated against long-term postoperative morbidity or mortality outcomes in veterinary medicine. These species-adapted criteria provide a necessary foundational framework for future large-scale investigations into canine neurobehavioral outcomes. Finally, the cohort consisted of large-breed, systemically healthy dogs undergoing TPLO procedures, representing a specifically orthopedic and somatic pain model. These findings may not generalize geriatric, pediatric, frail, or critically ill patients, nor to patients undergoing procedures involving severe visceral nociception.

5. Conclusions

In conclusion, this study demonstrates that incorporating continuous frontal EEG and PTA monitoring provides a more comprehensive assessment of the anesthetized canine patient than standard cardiovascular monitoring alone. Relying solely on autonomic vital signs can mask profound cortical depression, creating a clinical blind spot that frequently results in Double-Low states. While multimodal protocols utilizing systemic opioids induce the most profound and consistent cortical suppression, all evaluated protocols demonstrated the potential to disconnect brainstem-driven hemodynamic stability from cortical consciousness. Furthermore, transitioning rapidly from these deeply suppressed cortical states to wakefulness directly predisposes dogs to rough and dysphoric recoveries. Ultimately, modern veterinary anesthesia requires monitoring of autonomic balance and cortical integrity, in addition to conventional cardiorespiratory parameters, to optimize perioperative neurophysiological stability and improve postoperative recovery outcomes.