The effect of the hypertensive phase on the long-term outcomes of Ahmed glaucoma valve (AGV) implantation

To investigate the long-term effect of hypertensive phase (HP) on the clinical outcomes of Ahmed glaucoma valve (AGV) implantation. The records of patients with different etiologies of glaucoma who underwent AGV implantation with at least 3 years of follow-up were retrospectively reviewed. HP was defined as the IOP > 21 mm Hg during the first three months after surgery. The main outcome measure was cumulative success defined as 5 < IOP ≤ 21 mmHg and 20% reduction from the baseline with or without IOP lowering medications. Results that do not achieve cumulative success or undergo glaucoma reoperation during the follow-up period are considered failures. The secondary outcome measures were intraocular pressure (IOP) and the number of glaucoma medications. A total of 120 patients (28 patients of HP, 92 patients without HP) with an average age (± SD) of 48.9 ± 19.6 years and a mean follow-up of 4.5 ± 1.4 years were enrolled. The mean duration of survival was 5.3 ± 0.5 years in HP which was significantly shorter than 6.4 ± 0.2 years in non-HP (log rank = 4.2, P = 0.04). Mean IOP and number of IOP lowering agents were higher in postoperative visits at 1,2, 3, and 4 years in HP patients compared with non-HP (all Ps < 0.01). Higher baseline IOP was significantly associated with higher rates of surgical failure. In the long-term follow-up, the duration of survival was significantly longer in the non-HP group. In the non-HP group, the failure rate was significantly lower than the HP group.


3
Vol:. (1234567890) have an association with lower rates of success in comparison with trabeculectomy. [1,2] The hypertensive phase (HP) occurs in approximately 56 to 82% of those patients undergoing GDD implantation, so it is considered as a natural phase following the prior hypotensive phase observed in the first 3 weeks postoperatively [3,4]. The exact mechanism underlying the HP is poorly understood yet. However, the exaggerated process of wound healing around the implant still is the most acceptable mechanism [3]. The timing of the HP, starting over the first postoperative weeks, reaches a peak after 1 month, and is partially resolved over three to six months, supports the role of fibrosis formation and its maturation as major causes of the HP [5]. Egress of inflammatory mediators and cytokines from the aqueous humor to the subconjunctival space around the shunt is proposed as a stimulator of fibrous formation.
Higher baseline intraocular pressure (IOP) [6], the type and material of the device [7], the intrinsic characteristics of the devices like opening and closing pressure of the tube [8], and longer axial length [9] have been reported as the risk factors for the development of HP. Tsai et al. [10] reported a significantly higher mean IOP after one month in patients undergone Ahmed glaucoma valve (AGV) implantation compared to the patients who undergone Baerveldt glaucoma implantation (BGI). The results of Ahmed versus Baerveldt [11] study and Ahmed and Baerveldt comparison [12] study supported the finding of higher IOP in the AGV group after one month from the operation, while the early postoperative IOP on day 1 and a week later was lower in the AGV group. Similarly, the use of the Molteno device is associated with lower IOP compared with the AGV [13].
Considering HP as a natural phenomenon after GDD implantation was questioned by some authors in their studies. In this regard, Nouri-Mahdavi and Caprioli [14] demonstrated the worse IOP outcome and a higher need for glaucoma medication by passing 6 to 9 months from the AGV implantation. The authors hypothesized whether the HP is a spectrum of a failing AGV or is a transient phenomenon with no worse outcome. Multiple reports after Nouri-Mahdavi's study confirmed the worse success rates after the HP development [9,15]. However, to the best of our knowledge, the data on the long-term effect of HP on the surgical outcome are still limited.
The present study aimed to investigate the longterm effect of HP on AGV implantation outcome and the factors contributing to the development of HP. We hypothesized that HP could affect the long-term outcomes of AGV implantation.

Methods
The records of all glaucoma patients who undergone AGV implantation between Jan 2010 and Jan 2019 were retrospectively reviewed in the present study. All surgeries were performed in two tertiary glaucoma practice centers (Rassoul Akram and Imam Hossein Hospitals). The study protocol was approved by a local ethics committee and adhered to the tenets of the Declaration of Helsinki.
The patients who underwent AGV implantation and were followed up for at least 3 years were included. The exclusion criteria consisted of age less than 18 years and any non-glaucoma-related intraocular surgery within the follow-up period. HP was defined as follows: IOP > 21 mm Hg (with or without IOP lowering medication) within the postoperative 3-month period not resulted from tube obstruction or retraction. HP resolution was defined as an IOP < 22 mm Hg along with (1) a reduction in the IOP by ≥ 3 mm Hg with the same number of medications or less or (2) a reduction in at least 1 medication with the same IOP or less. According to the abovementioned definition, we categorized the included patients either in the HP group or in the non-HP group. Besides, if the HP has started within the first 5 weeks postoperatively, the HP was considered as early HP.

Surgical technique and postoperative visits
All surgeries were performed by two glaucoma specialists (NN, MY) using the uniform method for the surgery. Under general or local anesthesia, a fornixbased conjunctival flap was made in the superotemporal quadrant after priming the AGV (model FP7, New World Medical, Rancho Cucamonga, CA, USA), and the plate was then secured to the sclera with an 8-0 nylon suture, located approximately 10 mm behind the surgical limbus. The tube was then inserted into the anterior chamber using a 23-gauge needle. The anterior part of the tube was covered with a scleral patch graft (5*5 mm). Thereafter, the conjunctiva was secured with 8-0 Vicryl sutures. At the end of the surgery, we inferotemporally used subtenon methylprednisolone. Postoperatively, all the patients were prescribed ciprofloxacin eye drop 4 times per day for 10 days as well as betamethasone eye drop every 2 h for a 2-week period that was tapered over the next 8-12 weeks by considering the inflammation of the surgical area. The patients were then examined on a postoperative day 1, and afterward, weekly for 4 weeks, and finally, every 1-3 months, based on the clinical judgment. Glaucoma medications were started in a stepwise fashion when the target pressure was not achieved or when the HP was diagnosed. If no contradiction was present, aqueous suppressants were the first choice.

Data gathering
The following data were extracted from the patients' records: age, sex, stage, and type of glaucoma, preoperative and postoperative IOP, number of medications, intraoperative and postoperative complications, previous intraocular surgeries, and systemic diseases.

Statistical analysis
The primary outcome measures were cumulative success defined as 5 < IOP ≤ 21 mm Hg and at least 20% reduction compared to the baseline level with or without medications. Failure is considered when success definition is not addressed for at least 2 visits 3 months apart or glaucoma resurgery is necessary. The secondary outcome measures were IOP and the number of glaucoma medications. The normality of data was tested using Kolmogorov-Smirnov test and Q-Q plot. Mean was presented for variables with a normal distribution like IOP, median for variables without normal distribution like glaucoma medication, and frequency and standard deviation were used to describe the obtained data. To evaluate the changes in IOP and glaucoma medication at different intervals, linear mixed models were used. Moreover, to compare IOP and glaucoma medications between the groups, student T-test and Mann-Whitney were used, respectively. Complete success was defined as the intraocular pressure between 5 and 21 mmHg and 20 percent reduction compared to the baseline level without glaucoma medication. The qualified success was defined in the same way without considering the use of any glaucoma medication.
Kaplan-Meier survival analysis was used to assess the cumulative probability of success. All statistical analyses were performed using SPSS (version 25, IBM, Chicago, Illinois). A P-value less than 0.05 was considered statistically significant.

Results
The records of 120 included patients were reviewed in this study. The mean (± SD) duration of followup was 4.5 ± 1.4 years. The average age (± SD) of the study participants was 48.9 ± 19.6 (ranged from 18 to 88) years old. Twenty-eight (23.3%) patients experienced the hypertensive phase which was at 41.61 ± 29.03 days after surgery. IOP lowering agents also started in 19 patients with advanced glaucoma that required a lower IOP based on the targeted pressure between the second and the fourth weeks. The exact decision on the occurrence of HP was difficult in these patients. These patients were categorized as non-HP if the IOP was estimated to be lower than 22 after the adjustment for the effect of drugs. The patients who experienced the hypertensive phase were compared to the patients without HP in terms of age, sex, baseline IOP, baseline medication, and glaucoma type and previous glaucoma surgery ( Table 1).
Complete success (success without IOP lowering medications) occurred in 14.2% of the non-HP patients versus in 0 HP patients (P = 0.02, Table 2).
The mean IOP of the group with HP has significantly decreased from 29.6 ± 9.7 mmHg (mean number of meds = 3.07 or with a median number of medications = 3) at baseline to 19.3 ± 6.7 mmHg after 5 years with a median of 3 glaucoma medication. The mean of the reduction in IOP was 31% after 5 years. Correspondingly, an average IOP has significantly decreased from 27.4 ± 7.8 mm Hg with the mean number of 3.04 or with the mean number of medication = 3) at baseline to 16.4 ± 5.6 mm Hg after 5 years. The median number of glaucoma medications and the mean of the reduction in IOP were 2 and 36%, respectively (Table 3, Fig. 2). As shown in Table 4, the mean number of glaucoma medications was higher in the HP group compared to the group without HP at all intervals. In the fifth year, the median number of glaucoma medications in the HP and control groups was 3 and 2, respectively (Fig. 3).
The average of IOP in eleven patients experiencing early HP was 28.4 ± 3.9 mm Hg and in 17 patients in the late HP group, it was 28.9 ± 4.7 mm Hg (P-value: 0.4). In year 1, despite the use of glaucoma medications, IOP in 6 (22.2%) patients remained higher than 21. The mean recovery time in recovered HP was 9.44 ± 3.04 months. The reason for the failure of surgery in all cases was the lack of postoperative IOP necessary for success definition also 6 cases ( 3 cases in the HP group and 3 cases in the non-HP group) required resurgery for uncontrolled IOP.

Complications
Tube exposure occurred in 5 patients treated with conjunctival free graft. Five patients experienced choroidal detachment, of them 4 were conservatively managed and one underwent choroidal drainage. Corneal decompensation was observed in 2 patients who were conservatively managed. Two groups were compared in terms of overall complications (4 vs 11 in HP vs non-HP, Chi-square test, P = 0.74).

Discussion
The results of the present study show lower surgical success, higher IOP, and more glaucoma medications in patients experiencing HP after AGV implantation. Besides, the observed difference remained in the long-term period of 5 years, which is shown for the first time to the best of our knowledge.
Our results are in line with those of the previous reports investigating the effect of HP on the IOP outcome. Nouri-Mahdavi and Caprioli [14] found a higher mean IOP and more required glaucoma medications by passing 6 to 12 months from surgery in eyes with an HP. Similarly, Jung [9] reported higher IOP in the HP group compared to the non-HP group after 1 year, while the difference in surgical success did not reach a significant level. Subsequently, Jeong et al. [15] reported lower IOP control and worse success in eyes with HP after 2 years from the operation. Our study with a longer duration of follow-up has also shown a lower rate of success in the HP group.
Similar to Nouri-Mahdavi et Caprioli's study [14], we considered the IOP more than 21 mm Hg after 3 months, as a surgical failure, rather than an HP. Although the exact mechanism underlying HP has not been exactly clarified yet, various mechanisms have been proposed in this regard. The most plausible explanation relates the HP to the normal wound healing process, consequent fibrosis formation, and bleb encapsulation [3]. The histopathological studies investigating the wound healing process after GDD implantation reported a difference between the staged processes and procedures in which the aqueous flow is immediately established like AGV [16]. The mechanical pressure of the aqueous or the pro-inflammatory cytokines present in AC was hypothetically indicated to result in a full-thickness thick bleb wall around the plate [17]. The HP usually peaks after one month, which accords with the timing of fibrosis formation around the AGV plate. In contrast, Jung et al. [18] using anterior segment optical coherence tomography demonstrated thinner wall of the bleb accordant with the timing of HP and then postulated the collagen cross-linking as the mechanism underlying the pathophysiology of thinner bleb and HP. Moreover, the HP was not resolved in approximately two-third of the included patients. Besides, all HP patients needed glaucoma medication in the first year.
Moreover, there is a potential role for steroids as an inducer of high IOP after the surgery.
Correspondingly, Yuen et al. [19] reported lower HP rates in patients receiving nonsteroidal inflammatory drugs (NSAID), as a substitute for the topical steroid. Besides, the inflammatory cytokine level was found to be correlated with higher IOP in the glaucomatous eyes with an HP [20]. However, it still seems unclear whether these cytokines are stimulated with high IOP or they are contributing to the development of the HP. Intrinsic characteristics of the implanted device might also play a role in the HP since the HP more frequently occurs after AGV implantation compared to Baerveldt [11,21] or Molteno devices [13]. The valved nature of the AGV and lower surface area has been proposed as the possible cause of a higher prevalence of HP after AGV implantation.
In this study, we estimated the prevalence of HP lower than the reports in the literature [3]. The incidence of HP widely differs among different studies. The earlier studies reported higher prevalence rates (56% [14] and 82% [5]), while more recent studies estimated lower prevalence rates (31% [7,9]). The study by Won et al. [7] is similar to our study in terms of HP definition and AGV model, but has a shorter follow-up period. They found more HP than us (31.1% and 23.8%, respectively), but in their study there was no difference in overall success between the two groups, which seems to be due to the small difference in the definition of success with more rigor (A 20% reduction in IOP compared to the baseline we added in the definition of success). But similar to our study, the overall success rate in the non-HP group was significantly higher in the first year after surgery.
Because of the worse outcome after HP, multiple methods were proposed to prevent the HP including intraoperative and postoperative triamcinolone [22] or antimetabolite injection [23], early start of glaucoma medications [24], and medical suppression of aqueous humor soon after the surgery [17]. Pakravan et al. [24] reported the HP difference as 66.0% and 23.4% (P < 0.001%) in groups with the early start of fixed dorzolamide and timolol, respectively. In another study, intraoperative subtenon triamcinolone injection resulted in the HP difference between the two study groups (26% vs 52%, P = 0.027, respectively). The use of ketorolac in comparison with steroid eye drops has also led to lower HP occurrence ( 31% vs 53%, P = 0.27) [19]. However, we did not apply the specific proposed methods for the prevention of the hypertensive phase. The IOP lowering medications were started either when the IOP passed the target pressure or when there was a pattern of the continuous rising of IOP between follow-up visits.
Another interesting result of the present study is the worse outcome in the group experiencing early HP compared with the group of late-onset HP. Since the difference did not reach a statistically significant level, making a clear conclusion is not possible. However, in our experience, it was shown that HP entails a heterogeneous group of patients with different outcomes. Furthermore, Cheng et al. [8] investigated the opening and closing pressures in the AGVs and then hypothesized that high opening pressure (cutoff of 18 mm Hg) and low closing pressure (cutoff less than Mean ± SD 2.9 ± 0.9 3 (2, 4) 2.0 ± 1.3 2 (1, 3) < 0.001 P from baseline Fig. 3 Glaucoma medication in the group with the hypertensive phase and without hypertensive phase 7 mm Hg) were both related to the early hypertensive phase, while they did not affect the late HP. Therefore, we assume that there may be some tube-related factors affecting early HP that remain for a longer duration compared to late HP and consequently result in a worse outcome.
Our limitations are also included, but they are not limited to the retrospective nature of the study, small sample size, and different types of glaucoma.
In conclusion, we reported worse IOP outcomes, higher needs to glaucoma medications, and lower success rates in patients who experienced a hypertensive phase in the 5 years postoperatively. Patients experiencing HP in the earlier period after the surgery might also have a worse outcome compared to the patients experiencing HP for a period of 1.5 to 3 months after the procedure. In this regard, larger studies are required to investigate the possible difference between two groups of HP.
Funding No funding was received.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors declare no conflict of interest.
Ethical approval All procedures in this study were approved by the Ethics Committee of the Eye Research Center, Rassoul Akram Hospital, Iran University of Medical Sciences, Tehran, Iran.
Consent to participate Written informed consent was obtained from each participant.