Surgical Performance Determines Functional Outcome Benefit in the Minimally Invasive Surgery Plus Recombinant Tissue Plasminogen Activator for Intracerebral Hemorrhage Evacuation (MISTIE) Procedure

Abstract

BACKGROUND

Minimally invasive surgery procedures, including stereotactic catheter aspiration and clearance of intracerebral hemorrhage (ICH) with recombinant tissue plasminogen activator hold a promise to improve outcome of supratentorial brain hemorrhage, a morbid and disabling type of stroke. A recently completed Phase III randomized trial showed improved mortality but was neutral on the primary outcome (modified Rankin scale score 0 to 3 at 1 yr).

OBJECTIVE

To assess surgical performance and its impact on the extent of ICH evacuation and functional outcomes.

METHODS

Univariate and multivariate models were used to assess the extent of hematoma evacuation efficacy in relation to mRS 0 to 3 outcome and postulated factors related to patient, disease, and protocol adherence in the surgical arm (n = 242) of the MISTIE trial.

RESULTS

Greater ICH reduction has a higher likelihood of achieving mRS of 0 to 3 with a minimum evacuation threshold of ≤15 mL end of treatment ICH volume or ≥70% volume reduction when controlling for disease severity factors. Mortality benefit was achieved at ≤30 mL end of treatment ICH volume, or >53% volume reduction. Initial hematoma volume, history of hypertension, irregular-shaped hematoma, number of alteplase doses given, surgical protocol deviations, and catheter manipulation problems were significant factors in failing to achieve ≤15 mL goal evacuation. Greater surgeon/site experiences were associated with avoiding poor hematoma evacuation.

CONCLUSION

This is the first surgical trial reporting thresholds for reduction of ICH volume correlating with improved mortality and functional outcomes. To realize the benefit of surgery, protocol objectives, surgeon education, technical enhancements, and case selection should be focused on this goal.

ABBREVIATIONS

CT

computed tomography

EOT

end of treatment

IVH

intraventricular hemorrhage

Intracerebral Hemorrhage (ICH) is the most devastating type of stroke, affecting more than 2 million people annually.¹ The evolution of ICH management has yet to lead to concrete interventions significantly impacting functional outcomes.^2–4 Procedures aimed at reducing hematoma expansion have had only limited or no benefit on the ultimate functional outcome in this disease.^5,6 There has long been an interest in the surgical evacuation of ICH, with the aim of mitigating secondary effects of large-volume bleeds on the brain. However, recent well-conducted trials of craniotomy for supratentorial ICH failed to realize significant benefits in functional outcome.^7,8 Minimally invasive interventions for ICH evacuation have held more promise, presumably by sparing the brain added damage of surgical manipulations. But most studies to date have been small and the quality of evidence suboptimal by Cochrane criteria.⁹

One such technique, the minimally invasive surgery plus recombinant tissue plasminogen activator for ICH evacuation (MISTIE), was subjected to rigorous optimization in a Phase II trial.¹⁰ This led to a recently completed Phase III trial, randomizing 506 cases prospectively to best medical therapy per current guidelines with or without the MISTIE procedure, with blinded assessment of outcome at one year.¹¹ The procedure involved planning of the surgical approach and image-guided cannulation of an ICH through which initial aspiration of the hematoma is performed. A catheter is then secured in the remaining hematoma, which is then further dissolved and drained passively with serial doses of thrombolytic agent (Figure 1). Goals of the intervention were the reduction of the ICH volume to 10 to 15 mL at end of treatment (EOT) or until a maximum of 9 doses of alteplase (1 mg every 8 h) were administered. Each catheter is placed using image guidance along 1 of 3 trajectories depending on the anatomical location of the hematoma. These factors allowed standardization of the surgical task. Validation of the final catheter placement was assessed in real time by the trial's Surgical Center. Each catheter placement not receiving a passing grade required replacement, as further defined in the methods. From a surgical standpoint, the trial evaluated an explicitly defined technique that utilizes widely-available tools (Computed tomography [CT] scan, triage system, and image guidance). The trial midpoint analysis showed that surgical education and quality monitoring achieved consistent hematoma evacuation across sites worldwide, with surgeons at various stages of qualifications exceeding the performance previously noted in the Phase II trial.¹² Prior to the MISTIE trials there had been no rigorous standardization of surgical definitions or monitoring of the procedural task in minimally invasive evacuations of ICH, nor any reports addressing the generalizability of performance of such interventions.¹³

FIGURE 1.

A , Axial CT scan showing ICH of 37 mL with demonstrated stability. B , Subsequent MISTIE frontal catheter placed (trajectory A) by a surgeon qualified with probation*, with surgical center grading catheter placement as good*. C , Hematoma evacuated to final volume of 1 mL after 3 doses of alteplase and achieving mRS 2 after 1 yr. *See methods for detailed definitions.

The recently completed Phase III trial of this intervention was neutral on the primary outcome of modified Rankin Score (mRS) 0 to 3 at one year per intention to treat analyses.¹⁴ There was a benefit in mortality with no significant increase in severe disability (mRS 5). Prespecified secondary analyses identified a signal of possible greater benefit with deep hematomas and no apparent difference in benefit with earlier versus later execution of the surgical task (up to 72 h). The procedure was safe, with very low rates of, and no significant impact of surgery on, symptomatic brain bleeding or infections. Most notably, there was a significant functional benefit (mRS 0 to 3) in the 59% of cases achieving the prescribed EOT ICH volume ≤15 mL. We here present a per-protocol analysis of the impact of extent of ICH evacuation on functional outcome and factors associated with different thresholds of ICH evacuation in the cohort of the MISTIE III trial where the prescribed surgical task was deployed.

METHODS

Protocol and Patients

MISTIE (clinicaltrials.gov NCT01827046) is a first-of-kind phase III, surgical explanatory trial utilizing a prospective randomized open blinded end-point, which randomized 506 patients at 78 hospitals in North America, Europe, and Asia. The trial protocol has been published,¹¹ and it is detailed along with the prespecified statistical analysis plan on the trial website (Protocol Ver 4.0 http://braininjuryoutcomes.com/images/MISTIE3/MISTIE_III_Protocol_SAP_Ver4_14Apr15_final.pdf). It enrolled patients 18 yr of age or older with a recent (<24 h between symptom onset and diagnostic imaging) spontaneous, nontraumatic, supratentorial ICH with hematoma volume ≥30 mL, demonstrated stability of bleeding, no (or corrected) coagulopathy, and negative screening for vascular lesions or tumor etiologies. For inclusion, patients must have had an mRS of 0 or 1 prior to ictus, a Glasgow Coma Scale (GCS) of 14 or less or a National Institute of Health Stroke Scale of 6 or more at the time of presentation. Patients were evaluated at regular intervals for a full year after the hemorrhagic episode with independently adjudicated functional outcome as the primary endpoint as measured by mRS. The protocol was approved by each local institutional review committee as well as the consenting process prior to site activation and training. Other details of the training and qualification of surgical personnel, and surgical oversight by the trial's Surgical Center have been published, including details on patient consent.¹²

A total of 506 patients were randomized (between December 30, 2013 and August 15, 2017) with 255 assigned to undergo the MISTIE procedure and 251 assigned to the medical arm of the study. Of these, 1 patient was lost to follow up, 5 were considered ineligible and 1 had missing data, another 6 patients were excluded who underwent craniotomy (at 1, 3, 3, 32, 34, and 124 d) confounding assessment of the MISTIE procedure; leaving a total of 242 patients in the cohort for our analysis herein (Figure S1 in Supplemental Digital Content).

Thresholds of ICH Evacuation and Correlated Variables

Volumes of ICH were assessed independently by the Trial's centralized Reading Center, using volumetric software analysis of Digital Imaging and Communications in Medicine images of every CT scan uploaded by the trial sites.¹¹ The EOT ICH volume was defined on the CT scan performed 1 d after the last dose of alteplase, as required by the protocol. Three parameters were used to assess efficacy of hematoma evacuation: (1) absolute volume reduction, representing the stability scan ICH volume minus EOT ICH volume; (2) fraction of ICH evacuated, representing the ratio of absolute volume reduction to the stability volume; and (3) fraction of subjects reaching a target end-of-treatment volume of ≤15 mL. To assess homogeneity of surgical performance, these were compared in various surgical catheter approaches and in categories of surgeon qualification as at midpoint of the trial.¹²

The EOT ICH volumes were compared as a continuous variable, with the likelihood of achieving mRS 0 to 3 at 1 yr, or survival. Using the spline curve and best fit of this correlation, thresholds of 15 mL EOT and 70% extent of evacuation were identified for multivariate modeling. Since those thresholds or the mRS outcome could be confounded by initial ICH volume, associated intraventricular hemorrhage (IVH) volume, deep versus lobar ICH location, patient age, and initial GCS, those factors were included in multivariate models to assess the independent impact of ICH evacuation on the likelihood of mRS 0 to 3.

To assess whether further passive catheter drainage after last alteplase dose impacted outcome differently, the same analyses were conducted utilizing the final hematoma volume on CT scan obtained after MISTIE catheter removal (allowing varying periods of further passive drainage after last thrombolytic dose).

Patients were further subdivided into those achieving an EOT ICH volumes of ≤15, 20, and 30 mL, representing, respectively, the prescribed desired EOT end-point, and approximately the mean and 1 SD above the mean of EOT volumes. The 30 mL threshold was also the volume of ICH required for enrolling in the trial, hence residual ICH at that threshold would be an inarguable failure of the surgical task. Multiple variables were analyzed in order to determine which may be associated with reaching the 3 thresholds of ICH evacuation (list of all variables can be found in Table S1 in Supplemental Digital Content). Categories included relevant demographic factors, past medical history, hemorrhage features, hematologic characteristics at baseline, qualification levels and case experience by surgeon and site, catheter trajectories and placement, protocol deviations in two categories (site reported per pre-specified criteria and adjudicated by the trial's Safety Committee; and those in contemporaneous notes by the oversight Surgical Center personnel), and variables during the dosing period that could predispose to bleeding (elevated blood pressure readings and coagulopathy). These categories were analyzed in univariate correlations with the likelihood of reaching each of three thresholds of ICH EOT volume. Because of the number of variables queried and their potential co-correlation, significantly correlated univariate parameters were considered in multivariate modeling for association with reaching the same thresholds. Cases where the final EOT ICH volume was >30 mL also underwent independent review by three senior surgeons from the Trial's Surgical Committee (P.J.C., J-L.C., and M.R.H.) where the information gleaned from the case and all relevant contemporaneous Surgical Center notes were utilized in assessing root causes potentially contributing to suboptimal ICH evacuation.

Statistical Methods

Aspects of the statistical methods are further detailed in the Supplemental Digital Content.

RESULTS

Before randomization, mean ICH stability volume was 48.7 mL (n = 506). After the MISTIE intervention in the selected cohort (n = 242), the mean EOT ICH volume was 15.3 mL (SD 12.2), with 59.9% (n = 145) of patients achieving the predefined target EOT ICH volume of ≤15 mL. The mean percentage hematoma reduction was 69.7% and the mean absolute hematoma reduction was 34 mL (SD 14.7). Treatment duration, from randomization to EOT averaged 3.6 d (95% CI 3.4-3.7). Subsequent hematoma volume after catheter removal averaged 12.6 mL (SD 9.9), with a mean 35.5 mL (SD 15.2) removed (73.87% hematoma reduction) in 232 MISTIE patients who underwent post-catheter removal CT scans. Consistent with the mid-point analyses by Fam et al.,¹² hematoma evacuation measures including the frequency of reaching ≤15 mL EOT ICH volume were consistent throughout the trial (Figure 2).

FIGURE 2.

Volume reduction for each patient randomized to the MISTIE procedure expressed as percent evacuation. Efficacy of hematoma evacuation over the course of the trial as demonstrated by average hematoma evacuation rate and percent of subjects reaching target EOT volume of ≤15 mL. Cases that underwent craniotomy are denoted in red. Cases with negative percent evacuation included rebleeds and ICH expansions.

ICH Removal to ≤15 mL (or ≥70%) has Greater Likelihood of Achieving a mRS 0 to 3

Utilizing a linear spline model, reduction of ICH hematoma volume to ≤15 mL correlated with a good functional outcome, mRS 0 to 3 (OR 0.90, 95% CI 0.85 – 0.96, P = .002), suggesting a reduction beyond the 15 mL threshold increased the chance of having a good outcome by 10% for each additional mL of hematoma removed (Figure 3). Reduction above the 15 mL threshold did not significantly impact functional outcome (OR 0.99, 95% CI 0.96 – 1.02, P = .338). Similarly, reduction of hematoma volume by 70% or greater increased the chances of achieving an mRS 0 to 3. Each additional mL removed beyond 70% carried a 6% improvement in the chance of achieving a good outcome of mRS 0 to 3 (OR 1.06, 95% CI 1.02 – 1.10 P = .002). Percent reduction below the threshold of 70% did not impact the probability of achieving a good outcome (OR 0.99, 95% CI 0.97 – 1.01, P = .269) (Figure 4). This analysis was also run independently in those hematomas where the initial volume was ≤/ >45 mL, and the same trends were noted, but did not achieve significance (Figure S2A and S2B in Supplemental Digital Content).

FIGURE 3.

Cubic spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationship of hematoma reduction (EOT ICH Volume) to the probability of having a good outcome, mRS 0 to 3, at 1 yr. This is created by classifying dichotomized outcome as 1 or 0 (green dots at 1 = mRS 0 to 3, red dots at 0 = mRS 4 to 6). Further reduction beyond the 15 mL threshold (OR 0.09, P = .002) increased the chance of having a good outcome by 10% for each additional mL of hematoma removed (green shading showing statistically significant area of curve). Volume reductions to >15 mL threshold did not significantly impact the likelihood of achieving a good outcome.

FIGURE 4.

Cubic Spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationship of percent hematoma reduction (EOT ICH Volume) to the probability of having a good outcome, mRS 0 to 3, at 1 yr. This is created by classifying dichotomized outcome as 1 or 0 (green dots at 1 = mRS 0 to 3, red dots at 0 = mRS 4 to 6). Hematoma reduction beyond 70% increasing the chance of achieving a good outcome. Each additional mL beyond 70% carries a 6% improvement in the chance of achieving a good outcome (OR 1.06, P = .002) (green shading showing statistically significant area of curve). Percent reduction below the 70% threshold did not significantly impact the probability of achieving a good outcome.

We conducted the same analyses in correlation with survival at one year. Significantly better survival was accomplished with EOT ICH volume <30 mL (OR 5.545, CI 2.362-13.019, P <.001; with trends of benefit with even lesser removals). There was association with better survival with >53% hematoma removal (OR 3.896, CI 1.743-8.707, P ≤.001; Figure S3A and S3B in Supplemental Digital Content).

Furthermore, we analyzed those patients who underwent a CT scan post MISTIE catheter removal (n = 232) and showed that additional time and greater hematoma reduction with passive drainage, or the process of removing the catheter, had the same association with mRS 0 to 3 as was observed with EOT ICH volume analyses (Figure S4A and S4B in Supplemental Digital Content).

Multivariate models considering initial stability ICH volume, volume of associated IVH, deep vs lobar ICH location, presentation GCS, and age also revealed significant independent association of EOT volume ≤15 mL or ≥70% volume reduction with mRS 0 to 3 (OR 2.02 and 2.05, 95% CI 1.05 – 3.89 and 1.10 – 3.85, P = .035 and .025, respectively; Table 1).

TABLE 1.

Odds of Achieving 1-yr Functional Outcome mRS 0 to 3 Per EOT ICH Volume ≤15 mL and Per Percent Hematoma Removal >70% with Control for ICH Severity Factors

Model per EOT ICH volume mRS 0 to 3	OR	95% CI	P-value
EOT > 15 mL	REF
EOT Remain ≤ 15 mL	2.022	1.052 – 3.890	.035
Age < 56 yr	REF
Age 56 - < 67 yr	0.484	0.232 – 1.011	.054
Age ≥ 67 yr	0.097	0.037 – 0.252	< .001
GCS (3-8)	REF
GCS (9-12)	2.045	0.985 – 4.246	.055
GCS (13-15)	3.026	1.299 – 7.049	.010
Stability ICH < 35 mL	REF
Stability ICH 35 - < 45 mL	0.478	0.191 – 1.199	.116
Stability ICH 45 - < 55 mL	0.378	0.145 – 0.988	.047
Stability ICH ≥ 55 mL	0.260	0.010 – 0.679	.006
ICH lobar	REF
ICH deep	0.106	0.048 – 0.233	< .001
IVH volume (mL)	0.919	0.852 – 0.992	.030
Model per percent hematoma
removal mRS 0 to 3	OR	95% CI	P-value
EOT percent removed < 70%	REF
EOT percent removed ≥ 70%	2.053	1.095 – 3.848	.025
Age < 56 yr	REF
Age 56 - < 67 yr	0.493	0.238 – 1.020	.057
Age ≥ 67 yr	0.093	0.035 – 0.247	< .001
GCS (3-8)	REF
GCS (9-12)	2.134	1.029 – 4.423	.042
GCS (13-15)	2.986	1.262 – 7.065	.013
Stability ICH < 35 mL	REF
Stability ICH 35 - < 45 mL	0.483	0.195 – 1.208	.117
Stability ICH 45 - < 55 mL	0.336	0.127 – 0.887	.028
Stability ICH ≥ 55 mL	0.205	0.082 – 0.515	.001
ICH lobar	REF
ICH deep	0.100	0.045 – 0.225	< .001
Stability ICH volume (mL)	0.916	0.845 – 0.992	.030

Catheter Placement Accuracy, Surgeon Qualification Level, and Hematoma Evacuation

Table 2 compares each of the trajectories used for hematoma removal and surgeon qualification levels with catheter placement accuracy. Trajectory B had a significantly higher likelihood of initially poor catheter placement when compared to trajectories A and C (definitions in Figure S5 in Supplemental Digital Content). There were no significant differences in catheter placement accuracy rates among surgeons at different qualification levels (definitions in Table S1 in Supplemental Digital Content).

TABLE 2.

Catheter Placement Accuracy in Different Surgical Approaches and Among Different Surgeon Qualification Levels

		Catheter placement accuracy
	n (%)	Good	Suboptimal	Poor	RR*	CI	P-value
Trajectory A	115 (48)	77 (67)	34 (30)	4 (3)	0.51	0.18-1.49	.24
Trajectory B	47 (19)	17 (36)	22 (47)	8 (17)	2.22	1.01-4.87	.047
Trajectory C	80 (33)	55 (69)	20 (25)	5 (6)	0.90	0.34-2.36	.82
Surgeon prequalified	106 (44)	67 (63)	29 (27)	10 (9)	-	-	-
Surgeon qualified with probation	77 (32)	43 (56)	30 (39)	4 (5)	1.82	0.59-5.58	.40
Surgeon fully qualified	59 (24)	39 (66)	17 (29)	3 (5)	0.54	0.15-1.88	.38

(*) Relative risk of poor catheter placement is calculated by comparing each trajectory to the whole group. For surgeons' experience, surgeons with full qualification and qualification with probation were compared to prequalified surgeons as a reference.

Table 3 shows the efficacy of hematoma removal per trajectory used and per surgeon qualification. When comparing the 3 trajectories, trajectory C showed a lower likelihood of reaching the predefined trial end point of ≤15 mL of hematoma evacuation, compared to trajectories A and B. Surgeon qualification did not show a significant difference in any of the 3 metrics of absolute reduction, percent evacuated, or percent reaching ≤15 mL.

TABLE 3.

Efficacy of Hematoma Evacuation in Different Surgical Approaches and Different Surgeon Qualification Levels

	Trajectory A	Trajectory B	Trajectory C	P-value†	Surgeon prequalified	Surgeon qualified with probation	Surgeon fully qualified	P-value†
Number of patients (%)	115 (48)	47 (19)	80 (33)	-	106 (44)	77 (32)	59 (24)	-
Absolute reduction (mL)	34.9	30.9	33.8	.13	33.8	33.1	34.6	.63
Percent evacuated	71.6	69.0	67.2	.13	69.9	66.9	72.9	.12
Percent of subjects reaching ≤15mL	63.5	68	50	.027*	60.4	54.6	66.1	.25

^†Statistical analysis run for all combinations of variables, lowest P-value reported.

*Significant variable represents trajectory C as compared to A and B.

Factors Associated with Three Threshold of ICH Reduction

Patient distribution with three thresholds of ICH reduction is shown in Figure 5. Threshold groups were defined as:

1. Successful MISTIE evacuation to ≤15 mL, protocol goal (n = 145, 59.9%)

2. Evacuation to ≤20 mL, short of protocol goal (n = 182, 75.2%)

3. Evacuation to ≤30 mL, threshold of MISTIE trial enrollment (n = 216, 89.2%)

FIGURE 5.

EOT ICH volumes for all MISTIE enrolled patients, excluding craniotomy (n = 6). EOT of ≤30 mL were achieved in 216 cases (89.2%), ≤20 mL in 182 cases (75.2%), and ≤15 mL in 145 cases (59.9%).

Each specified variable (Table S1 in Supplemental Digital Content) was first analyzed in a univariable model with the likelihood of achieving the three respective thresholds. The significant variables in univariate correlations are noted in Table S2 in Supplemental Digital Content. These were incorporated into a multivariable model to assess their independent contributions (Table 4). Factors significantly and independently associated with EOT volume of >15 mL were the absence of history of hypertension, higher initial ICH volume, greater number of alteplase doses administered, irregular-shaped hematoma, surgical center protocol deviation, and catheter manipulation problem. The variables associated with EOT volume >20 mL were a higher initial ICH volume and greater number of alteplase doses administered, an irregular-shaped hematoma, surgical center protocol deviation, and catheter manipulation problem. The variables associated with achieving an EOT volume of >30 mL included greater initial hematoma volume and alteplase doses, catheter manipulation problems, and a catheter tract hemorrhage.

TABLE 4.

Variables Significantly Associated with Achieving EOT ICH Volumes at 15-, 20-, and 30-mL Thresholds (Multivariate Analyses)

Variables (no vs yes)	OR	95% CI	P-value
EOT 15 mL
History of hypertension	0.207	0.057 – 0.745	.016
Initial volume at stability scan	0.968	0.945 – 0.991	.008
Irregular-shaped hematoma	28.741	1.305 – 633.187	.033
Surgical center protocol deviation	74.580	3.292 - > 1000	.007
Catheter manipulation problem	107.454	4.751 - > 1000	.003
Number of alteplase doses	0.779	0.675 – 0.899	.001
EOT 20 mL
Initial volume at stability scan	0.955	0.926 – 0.984	.002
Irregular-shaped hematoma	123.803	4.946 - > 1000	.003
Surgical center protocol deviation	159.928	16.079 - > 1000	<.0001
Catheter manipulation problem	126.644	16.507 – 971.649	<.0001
Number of alteplase doses	0.644	0.521 – 0.797	<.0001
EOT 30 mL
Initial volume at stability scan	0.927	0.897 – 0.958	<.0001
Catheter manipulation problem	6.014	1.184 – 30.548	.031
Catheter tract hemorrhage	17.675	3.114 – 100.333	.001
Number of alteplase doses	0.699	0.559 – 0.874	.002

Surgeon and Site Case Experience and ICH Evacuation Efficacy

In addition to the qualification categories evaluated above, surgeon and site experience throughout the trial were assessed as a function of how many previous procedures had been performed up until the current procedure (Figure S6 in Supplemental Digital Content). While there was no significant difference in the surgeon and site experience among groups reaching different EOT volume thresholds, there was a threshold of 4 prior MISTIE trial cases by the surgeon and 7 prior cases by the site, above which there was no case with poor (>30 mL) EOT ICH (Figure 6A and 6B).

FIGURE 6.

A, Surgeon andB, site experience depicting the number of prior cases performed in the MISTIE trial until the date of the index case, in cohorts where EOT ICH volume was ≤ or >30 mL (significant variable in univariate analysis). (Whiskers represent the 3rd and the 1st quartiles with box line representing the median). No surgeon with experience >4 MISTIE procedures or a site with >7 MISTIE procedures had a patient with >30 mL EOT ICH volume. This difference was present at the 30 mL EOT volume threshold, but not at 20 or 15 mL EOT thresholds (not shown).

Illustrative Cases of Suboptimal ICH Evacuation

Root cause analyses of individual cases with residual EOT ICH volume >30 mL were reviewed by three independent senior surgeons. These endorsed likely contributive factors including catheter track hemorrhage, very large hematomas (mostly lobar), irregular-shaped multifocal (satellite) residual hematomas, and catheter manipulation problems and surgical protocol deviations noted contemporaneously by the Surgical Center. These are the same factors identified independently in multivariate statistical analyses. Case examples of suboptimal ICH evacuation and their root causes are illustrated in Figure 7. Frequencies of protocol deviations with different thresholds of evacuation are listed in Table S3 in Supplemental Digital Content. We list protocol deviations in patients not achieving 15 mL EOT in Table S4 in Supplemental Digital Content.

FIGURE 7.

Three example scenarios of failure to achieve the desired surgical evacuation. In Scenarios 1 and 2, surgical experience and protocol adherence could have achieved the desired outcome. In Scenario 3, inherent limitations related to the type of hematoma prevented the desired outcome, despite optimal surgical performance and protocol adherence. A and B , Scenario 1. A , Initial catheter placement was suboptimal (purple outline of catheter perforations and hematoma interface) but deemed suitable for dosing. B , After 9 doses of alteplase, there was 72.6 mL residual hematoma eccentric to the catheter. Replacing the catheter during dosing was suggested to the site in contemporaneous notes, but was not pursued. Instead, the site investigators administered all 9 doses through the eccentric catheter location despite evidence of suboptimal clearance. C and D , Scenario 2. C, Catheter was dislodged before completion of dosing. D, Surgical Center query could not verify that the catheter was properly secured at three points, as mandated in the protocol and emphasized in surgical education module/webinars (arrows highlight optimal catheter fixation in another case, arrowhead shows proper use of stopcock and star denotes proper labeling of MISTIE from EVD catheter). The site team aborted the procedure and declined replacing the catheter. Scenario 3. E and F: E, Despite excellent catheter placement, F, there was 24.1 mL residual hematoma in fragmented satellite configuration not amenable to further evacuation.

DISCUSSION

ICH evacuation has been the holy grail of ICH surgical therapies, yet no prior studies have correlated the success of the procedure with residual ICH volume. The best recently conducted pragmatic trials of open craniotomy failed to demonstrate a benefit for initial surgical management, but they did not analyze outcomes by postoperative residual or recurrent ICH volume, nor consider the possibility that better outcome could have been achieved in cases where the ICH was in fact removed.^7,8 Our findings in this explanatory trial, with strict oversight and monitoring of the process and outcomes of the surgical procedure, demonstrate that a threshold exists (of ≤15 mL EOT ICH or ≥70% hematoma evacuation) associated with favorable functional outcome, after controlling for disease severity variables. Mortality benefit was achieved with lesser extent of removal (30 mL). This should henceforth be considered when assessing results of surgical interventions for ICH. Smaller trials with less invasive techniques often report percent hematoma removal as proof of concept, but do not consider its association with outcome.¹³ MISTIE is a procedure that effectively and safely allows reduction of hematoma volume but the optimal endpoint threshold was never previously defined. A similar finding of better outcome with more efficient blood thrombolytic clearance of IVH was also observed in the CLEAR trial.¹⁵

Multivariate analyses revealed that a prior history of hypertension, a lower initial ICH volume, and fewer alteplase doses were associated with improved hematoma evacuation. Initial GCS did not affect this. This is intuitive from the perspective that most patients with hypertension were likely being concurrently treated with oral medicines or had deeper ICH locations. Smaller hematoma size already confers an advantage as these are nearer to goal. Fewer alteplase doses reflect the ease of hematoma lysis and responsiveness to alteplase, while more alteplase doses are needed in cases with less efficient clearance. Additionally, irregular-shaped hematomas were significantly associated with less efficient ICH removal. Special approaches to larger lobar hematomas might include plans for more than one catheter. These were allowed per the trial protocol but were seldom deployed. Replacing the catheter to allow better clearance of eccentric residual hematomas was another strategy often suggested by the Surgical Center overseeing team (Figure 7, Scenario 1), but was rarely used by the treating teams, who articulated lesser equipoise about risk-benefit of such maneuvers, particularly when they felt they had accomplished “good enough” evacuations. These decisions were often made jointly by the site principal investigator, stroke and neurocritical care teams and the surgeon of record, and reflected the teams’ collective biases about safety and risk-benefit in the absence of a clearly known threshold EOT for success. It was reassuring that volumes of residual ICH averaged ≤15 mL throughout MISTIE III and extent of removal exceeded 70% on average. This was reported at midpoint of the trial¹² and considered a major improvement in comparison to MISTIE Phase II performance.¹⁰ Until results of MISTIE III were analyzed, the effects of additional hematoma evacuation on functional outcome were unknown and we did not realize that these were not “good enough”. Average performance did not drive individual functional benefit from the procedure, when more than 40% of cases did not reach the 15 mL EOT threshold.

Protocol deviations recorded contemporaneously by Surgical Center oversight personnel correlated with less optimal ICH evacuation. Catheter manipulation/dosing problems were also significant factors (Figure 7, Scenario 2). Catheter track hemorrhages were also independently associated with >30 mL EOT ICH. These were often noted after the fact and did not prevent the suboptimal outcome in the individual case. They also frequently reflected the site team's decision not to pursue more aggressive hematoma evacuation endpoint per their individualized assessment of risk-benefit, as noted above. These factors were often not reported by the site and were not included in the separately tracked adjudicated major protocol deviations, illustrating an advantage of real time monitoring of surgical performance, particularly in the setting of evolving experience with the surgical task. In all, about a quarter of cases not reaching the 15 mL EOT goal involved some protocol violation that could potentially be mitigated with better education and experience.

We incorporated “Lessons Learned” in cases of suboptimal ICH evacuations or other technical issues in monthly Executive Committee meetings and included these in required surgical education webinars during the course of the trial (https://slideplayer.com/slide/10395752/). These included the importance of stabilizing any new bleeding, the need for fiducials and more lateral entry points and wider burr holes for posterior approaches, and better tips for securing the catheter and avoiding dislodging. These may have contributed to the exceptional safety performance in MISTIE III and to uniform hematoma evacuation performance by a large group of surgeons with limited experience with the procedure. Importantly, there were signals that greater experience with the surgical task may contribute to avoiding suboptimal performance. Using our threshold analysis, a clear delineation of >4 MISTIE procedures by the surgeon in the trial (or 7 procedure by the site) reliably avoided cases with only >30 mL EOT. This is similar to learning curve thresholds achieving acceptable outcomes in other interventional and surgical procedures.^16–18 In carotid artery stenting trials, posthoc analysis later revealed that the initial credentialing system was somewhat underpowered to what was needed to overcome the learning curve of the procedures.¹⁹ While the procedure itself deploys techniques learned by every neurosurgeon, clarification of requisite EOT goals and technical nuances of the protocol and increasing experience will likely enhance future performance.¹⁶ Surgeons would likely pursue more aggressive hematoma evacuation with the MISTIE procedure, by replacing catheters or being more aggressive with additional doses of alteplase, now that they are aware of the required thresholds for impacting outcome. It is possible that advanced technical adjuncts might also contribute to this objective.^20–27

Other MIS techniques are being developed, aiming to enhance the efficiency of thrombolytic delivery and hematoma clearance,²⁸ or deploying endoscopic tools²⁹ or various hematoma access and evacuation devices.^30–32 While these may promise more efficient or facile ICH evacuation, they need to be assessed with similar rigor to the MISTIE benchmarks, and should address whether the same generalization and efficiency considerations may influence functional outcome. These techniques and open surgery may have similar limitations in irregular-shaped hematomas or those with satellite residuals (Figure 7, Scenario 3).

Limitations

Observations and conclusions in this study are limited by the posthoc exploratory nature of our hypotheses. Statistical associations do not prove causation, and per-protocol methods introduce biases avoided in the primary intention-to-treat analyses. We also did not assess clot consistency vs the efficiency of removal. Nevertheless, we applied herein the most rigorous cohort analysis methodologies, including stringent multivariate controls, and we generated novel hypotheses about this procedure in light of results of the trial. Still, the EOT ≤15 mL endpoint or ≥70% evacuation efficiency must be tested pragmatically before we can apply the procedure in the community with best level and class evidence.

CONCLUSION

While mortality benefit can be achieved with lesser ICH removal, reduction of ICH to ≤15 mL EOT volume or ≥70% evacuation was needed for favorable functional outcome at 1 yr. This is the first description of specific thresholds of hematoma evacuation to impact mortality or functional outcome in ICH. Generalization of best performance with this procedure, or other techniques of this kind, will require focused surgeon education, emphasizing technical nuances, better demonstrated experience, and strict definitions of the task benchmark of success. Future studies should focus on methods to best achieve this target and may require individually tailored case selection for various techniques.

Disclosures

The sponsor/funding agency is: National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health (NIH), 9000 Rockville Pike, Bethesda, Maryland 20 892. Study Intervention Provided by: Genentech, Inc. (US and Canada only), 1 DNA Way, Mailstop 258A, South San Francisco, CA 94080. This trial is registered with ClinicalTrials.gov NCT01827046.

This material was presented as an abstract at the American Heart Association's International Stroke Conference on February 7, 2019 in Honolulu, Hawaii.

Neurosurgery Speaks! Audio abstracts available for this article at www.neurosurgery-online.com.

Supplemental Digital Content. Supplemental Tables and Figures. Figure S1. Population of MISTIE trial patients for surgical performance analysis. *Cases excluded: A total of 6 patients underwent craniotomy within 1 yr of randomization (3 were without hematoma evacuation, 1 was with hematoma evacuation, and 2 were craniectomies with hematoma evacuation). One was lost to follow up. A total of 5 were considered ineligible as adjudicated by the Data Safety and Monitoring Board (2 had ICH size that did not meet entry criteria of ≥30 mL, 1 had consent withdrawn, 1 had not received the MISTIE treatment, and 1 was not eligible as the etiology screen revealed an arteriovenous malformation). One had missing data. Figure S2 (A and B). Cubic Spline regression analyses (blue lines) showing the relationships of hematoma reduction (EOT ICH Volume) to the probability of having a good outcome, mRS 0 to 3, at 1 yr for cases with A, ≤45 mL and B, >45 mL starting volume (median of surgical arm) measured at stability scan. This is created by classifying dichotomized outcome as 1 or 0 (green dots at 1 = mRS 0 to 3, red dots at 0 = mRS 4 to 6). There is a trend towards beneficial effect with lower EOT volumes in both larger and smaller initial hematomas, but these were not statistically significant. Caution about interpreting nonsignificant segments of the plots with few contributing cases. Figure S3 (A and B). A, Cubic spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationships of hematoma reduction (EOT ICH Volume) to the probability of survival. A threshold at 30 mL shows that reduction to this level and beyond confers a survival benefit (OR 5.545, CI 2.362-13.019, P <.001). B, Cubic Spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationships of percent hematoma reduction (EOT ICH Volume) to the probability of survival. A threshold at 53% shows that reduction to this level and beyond confers a survival benefit (OR 3.896, CI 1.743-8.707, P <.001). These are created by classifying dichotomized outcome as 1 or 0 (green dots at 1 = living, red dots at 0 = deceased). Caution about interpreting nonsignificant segments of the plots with few contributing cases. Figure S4 (A and B). A, Cubic spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationships of hematoma reduction (EOT ICH Volume) to the probability of survival. A threshold at 30 mL shows that reduction to this level and beyond confers a survival benefit (OR 5.545, CI 2.362-13.019, P <.001). B, Cubic Spline regression analyses (blue line) and linear spline regression analyses (black line) showing the relationships of percent hematoma reduction (EOT ICH Volume) to the probability of survival. A threshold at 53% shows that reduction to this level and beyond confers a survival benefit (OR 3.896, CI 1.743-8.707, P <.001). These are created by classifying dichotomized outcome as 1 or 0 (green dots at 1 = living, red dots at 0 = deceased). Caution about interpreting nonsignificant segments of the plots with few contributing cases. Figure S5. Trajectory A: anterior approach through the forehead used for deep-seated ICH in the basal ganglia. This was used in 48% of MISTIE cases. Trajectory B: Posterior approach with entry point through the parietal-occipital area, frequently several centimeters lateral from the midline to avoid the occipital ventricular horn, used for deep-seated ICH occupying the posterior basal ganglia or thalamus. It was used in 19% of cases. Trajectory C: used for lobar hematomas, with entry point through the superficial area closest to the hematoma. Used in 33% of cases. In all approaches, catheter trajectory should optimally be along the longitudinal axis of the clot, and A and B, traversing its epicenter, although C, this was frequently difficult to achieve in several lobar hematomas because of their shape. Fam et al, surgical performance in minimally invasive surgery plus recombinant tissue plasminogen activator for intracerebral hemorrhage evacuation phase III clinical trial, Neurosurgery, 2017, 81, 5, 860 to 866, by permission of the Congress of Neurological Surgeons. Figure S6. Histogram showing surgeon (black bars) and site experience (red bars) indicating the number of MISTIE procedures completed at the time of the index procedure. Table S1. Variables Queried for Association with Reaching EOT Volume Thresholds (all categorical unless noted as continuous). Table S2. Variables significantly associated with likelihood of achieving EOT ICH volumes at 15, 20, and 30 mL thresholds (univariate analyses). Table S3. Frequencies of variables categorized by EOT ICH volume thresholds (15, 20, 30 mL). Table S4. Surgical center protocol deviations noted in cases not meeting the 15 mL EOT threshold.

COMMENT

The authors present the results of the surgical treatment arm of the MISTIE III study and analyze the impact of the extent of clot evacuation on functional outcome as determined by the modified Rankin scale score (mRS) at 1 year. While the overall trial results are under review, a brief summary is presented in the manuscript.

The trial randomized 506 patients in 78 hospitals to best medical therapy alone (251) or with the MISTIE procedure (255) involving image-guided cannulation of the hematoma with a catheter, initial aspiration with a 10cc syringe “until finding resistance”, injection of 1 mg of tPA q8h through the catheter 6 hours after placement (up to 9 doses), and passive drainage. The primary outcome measure in the intention to treat analysis was an mRS of 0−3 at 1 year and per the authors report it was neutral despite providing a mortality benefit. Deeper hematomas were reported to have greater benefit and time to intervention within the study window of 72 h did not favor earlier evacuation.

Surgical success was defined as achieving an end-of-treatment (EOT) volume of <15cc or a >70% clot evacuation. Mean overall hematoma volume was 48.7 mL and mean EOT volume of the 242 surgical patients analyzed was 15.3 mL, resulting in an evacuation rate of 69.7%, although only 59.9% of patients reached the EOT volume goal. A hematoma reduction beyond 15 mL correlated with a greater likelihood of achieving an mRS 0−3 (odds ratio, 0.9), which increased by 10% for each additional mL removed. Similarly, a hematoma evacuation >70% also increased the likelihood of achieving an mRS 0−3 (odds ratio, 0.99).

While evacuation rates surpassed those seen in MISTIE II, not reaching an EOT volume on more than 40% of patients raises questions about the efficacy of the procedure and leaves us wondering if a technique resulting in greater clot reduction could have resulted in a greater difference in mRS scores at 1 year. Nonetheless, these findings will define treatment goals for all ICH trials to come and provide, for the first time, class I evidence of the impact of surgical performance on clinical outcomes in this disease.

Fully qualified surgeons achieved the EOT volume 78% of the time, while the “qualified with probation” or “prequalified” achieved it 54.6% and 60.4% respectively. Once a surgeon reached 4 cases or the site reached 7 cases an EOT volume >30 mL did not occur. Although the surgical protocol does not require microsurgical technique, experience with frameless stereotactic guidance for biopsy, tumor resection, ventriculoperitoneal shunt placement, and other similar procedures could affect surgeon and site performance. Future trials should incorporate sufficient training and objective performance metrics for the technique being tested prior to receiving approval to enroll their first patient. While technical adjuncts such as ultrasound, magnetic-guided catheter placement, CT/MR guided targeting and others may help improve accuracy, these should be part of the protocol and the site/surgeon should demonstrate proficiency in those techniques as well.

Overall, MISTIE III constitutes a well-designed and rigorously conducted trial that will significantly influence the field. It will serve as the foundation for the design and execution of the next generation of ICH trials evaluating exciting new technologies and highlights the communal enthusiasm for finding a surgical solution for this disease.

Gustavo Pradilla

Daniel L. Barrow

Atlanta, Georgia