Comparison between intravenous chemotherapy and intra-arterial chemotherapy for retinoblastoma: a meta-analysis

Background Intravenous chemotherapy (IVC) and intra-arterial chemotherapy (IAC) have become the primary treatments for retinoblastoma; however, some controversy remains over which method is more effective. We conducted a meta-analysis to compare the clinical efficacy of IVC and IAC. Methods We systematically searched literature published on PubMed, Embase, and Cochrane Library up to May 2017. Studies containing either IAC or IVC that reported on efficacy were included. The effects estimate was expressed as a pooled rate with 95% confidence interval (CI), using a fixed-effects or random-effects model. Results Twenty-six studies were identified which included 1541 eyes (IAC: 11 trials, 445 eyes; IVC: 16 trials, 1096 eyes). The mean follow-up times were 49.4 months (range, 13.0–105.3 months) for IVC and 21.7 months (range, 8.8–38.7 months) for IAC. For the International Classification of Intraocular Retinoblastoma (ICRB) grading, the overall success rate was higher with IAC than with IVC (75.7% [95%CI: 65.7%–83.6%] vs. 69.5% [95%CI: 51.9%–82.8%], P < 0.001). The globe salvage with IAC was higher than with IVC in group D eyes (79.5% [95%CI: 71.8%–85.4%] vs. 55.1% [95%CI: 45.6%–64.2%], P < 0.001), but not in groups B (95.8% [95%CI: 57.5%–99.7%] vs. 82.5% [95%CI: 58.9%–94.0%], P = 0.163), C (91.3% [95%CI: 65.9%–98.3%] vs. 89.0% [95%CI: 69.0%–96.7%], P = 0.212), and E eyes (51.2% [95%CI: 37.0%–65.2%] vs. 43.2% [95%CI: 18.3%–72.1%], P = 0.578). IAC and IVC were not significantly different regarding the recurrence and metastasis rates (15.0% vs. 15.4%, P = 0.148 and 2.7% vs. 0.6%, P = 0.194, respectively). For Reese-Ellsworth (RE) grading, IAC had a higher globe salvage in groups IV (90.9% [95%CI: 56.0%–98.7%] vs. 66.3% [95%CI: 32.4%–89.0%], P = 0.047) and V eyes (83.2% [95%CI: 72.0%–90.5%] vs. 59.9% [95%CI: 43.1%–74.6%], P = 0.003), but not in group I-III eyes (88.6% [95%CI: 58.3%–97.7%] vs. 88.1% [95%CI: 76.6%–94.4%], P = 0.244). The overall success rate was higher in IAC than in IVC (87.1% [95%CI: 78.1%–92.7%] vs. 77.3% [95%CI: 68.1%–84.4%], P = 0.033). Conclusions IAC may be superior to IVC for the treatment of retinoblastoma, with a higher overall success rate and higher globe salvage in group D or groups IV and V eyes. Electronic supplementary material The online version of this article (10.1186/s12885-018-4406-6) contains supplementary material, which is available to authorized users.


Background
Retinoblastoma is the most common intraocular malignancy of childhood (approximately 1/15,000-20,000 live births) and it accounts for 4% of all pediatric cancers [1]. Two-thirds of all cases of retinoblastoma are diagnosed before the age of two years. All patients with bilateral retinoblastoma and approximately 10%-15% of children with unilateral disease carry a germline mutation which is transmissible to their off spring [2]. In developed countries, the survival rate approaches 98%; however, due to the limitations of health care in low-income countries, it is much lower at 40% [3,4]. Before the 1990s, retinoblastoma was mainly treated using enucleation and external beam radiotherapy (EBRT). However, these methods are associated with numerous complications, including loss of vision and severe toxic side effects. Currently, firstline conservative management of retinoblastoma has moved from EBRT and enucleation to intravenous chemotherapy (IVC) or intra-arterial chemotherapy (IAC), consolidated by focal treatment. The therapeutic efficacy of IVC and IAC has received increasing attention.
IVC was first used in 1953 [5]. The standard protocol is the vincristine-etoposide-carboplatin (VEC) program used as a combination triple-drug therapy, typically performed for six cycles [6]. VEC may yield better results in eyes with advanced cancer when combined with local therapeutic methods. The detrimental effect of IVC on children is worth noting. Since eyes with vitreous or subretinal seeding are less sensitive to IVC treatment, IVC may not perform well in cases of advanced tumors. In addition, IVC is associated with certain adverse reactions: carboplatin is known to cause ototoxicity and etoposide has the potential to cause acute lymphoblastic leukemia. IAC was first described by Reese and colleagues in 1950 [7]. The technique was pioneered in Japan [8] and was later popularized by Abramson et al. [9]. The procedure involves directly injecting concentrated doses of chemotherapeutic drugs (melphalan, topotecan, or carboplatin) into the ophthalmic artery using a modern microcatheter (Fig. 1), to increase the concentration of chemotherapy drugs 10-to 30-fold at the tumor site. Consequently, the concentration of the drugs in the peripheral blood is minimal. Each eye requires an average of three treatment cycles and each cycle is planned at a 4-week interval. Successful treatment is indicated by a decrease in the size of the tumor. Remaining tumors can be eliminated using laser, cryotherapy, or radioactive plaques. Previous studies have reported the efficacy and safety of this approach [10][11][12][13]. However, due to the high concentration of the chemical drugs that are used in IAC, local complications are very high. Thus, the issues around improving globe salvage as well as reducing the risk of local complications must be addressed urgently.
At present, both IVC and IAC are first-line treatments for retinoblastoma in the clinical setting. Although IAC has been shown to have excellent therapeutic effects, whether it can replace IVC remains controversial. Some single-arm studies have reported either on IAC or IVC, but only one study compared these two methods in unilateral group D retinoblastoma [14]. Moreover, the outcomes of studies that evaluated success rates of these methods have been discrepant due to the varying quality of the research and the different sources and capacities of the samples. Hence, we performed this study to compare the clinical use of IAC and IVC to identify which method is more effective or to determine if both are necessary for the treatment of retinoblastoma. The improved quality of evidence resulting from this meta- Fig. 1 The technique of intra-arterial chemotherapy (IAC) treatment. a: An arteriogram was performed to indicate the takeoff of the ophthalmic artery from the internal carotid artery. b: Using fluoroscopy and roadmap guidance, the microcatheter selectively catheterized the ophthalmic artery. c: As soon as the microcatheter was in a stable position at the ostium of the ophthalmic artery, then pulse-inject drugs analysis is expected to be helpful for doctors in their clinical practice.

Search strategy
PubMed, Embase, and Cochrane Library databases were searched without geographical, publication type, and language restrictions for literature published up to May 2017. The literature was searched by combining Medical Subject Headings (Mesh) and multiple free words. Mesh and free words were combined using a logical 'OR' operator. The results of the Mesh and free words search were combined using a logical ' AND' operator. As the search terms and strategy were not perfect, we also used references from papers to supplement the results.

Inclusion criteria
Two authors (LZ and YHD) reviewed the articles by reading the title, abstract, and full text independently; a third author (XKM) resolved disagreements and recorded the excluded study as well as the reasons for its exclusion. Inclusion criteria were as follows: (1) only IAC or IVC as the primary treatment method (not used in combination); (2) eyes were grouped according to the International Classification of Intraocular Retinoblastoma (ICRB) system, or the Reese-Ellsworth (RE) classification system; (3) studies that provided efficacy data; (4) more than 10 eyes included in a study; and (5) when we encountered multiple publications of the same clinical trial, we selected the latest or most complete publication. We excluded studies if they were published as reviews or case reports.

Classification of retinoblastoma
Classification of retinoblastoma is necessary to formulate appropriate management strategies and improve predictability of the treatment outcomes. Formerly, RE grading was widely used; however, it was later replaced by ICRB system which included the Philadelphia version and the Children's Hospital Los Angeles (CHLA) version [15]. Therefore, RE grading are mainly reported in earlier articles while ICRB system are mainly reported in articles published in recent years. In order to include valuable data as much as possible, we selected RE grading and ICRB system as the basis for grouping. Fig. 2 shows the fundus photographs of groups A to E eyes.

Data synthesis and analysis
Meta-analysis was performed for an outcome only if there were appropriate data for at least two single arms in each group. The rates of different groups were used as weighting variables and expressed as a pooled rate with 95% confidence intervals (CI). We evaluated heterogeneity across studies using I 2 statistics, with values of > 25%, > 50%, and > 75% representing mild, moderate, and grievous heterogeneity, respectively. If I 2 > 50%, a random-effects model was applied in case of statistical heterogeneity, otherwise a fixed-effects model was used. We used the Chi-square test or Fisher's exact test to compare the rates of different groups. A P-value of < 0. 05 was considered statistically significant. Statistical analyses were performed using Comprehensive Meta-Analysis version 2 and SPSS version 16.

Primary endpoints
For ICRB grading, the comparison in group A eyes could not be performed due to the absence of data from IAC. Although no significant differences were observed in the globe salvage rates between IAC and IVC in group B eyes (95.8% [95%CI: 57.5%-99.7%] vs. 82.5% [95%CI: 58.9%-94.0%], P = 0.163) and group C eyes (91.3% [95%CI:   [20], whereas it occurred in six eyes in IAC [29,38]. However, no evidence supported that IVC had an advantage over IAC (2.7% vs. 0.6%, P = 0.194). In a study by Turaka et al. [39], 4% (6/156) of patients with germline retinoblastoma treated with IVC as first-line therapy developed SMNs. These SMNs included osteosarcoma (n = 3), tectal glioma (n = 1), acute promyelocytic leukemia (n = 1), and one patient with both rhabdomyosarcoma (temporal fossa) and conjunctival/orbital malignant melanoma. Shields et al. [25] reported one case of osteosarcoma. However, no SMNs were reported in patients treated with IAC. For RE grading, recurrence occurred in only five eyes (two trials). No tumor metastasis or SMNs was reported.
The details of adverse events of IAC (seven trials) and IVC (nine trials) are listed in Table 4. Adverse events included systemic complications and ocular complications. IAC had more systemic complications than IVC, including neutropenia and infections. Ocular complications were also reported more frequently in IAC, which included vascular injury, spasm, obstruction, and a series of related organ ischemic lesions.

Discussion
The current commonly used treatments for retinoblastoma include IVC and IAC, along with consolidated therapy based on tumor staging. For ICRB grading, we found a significantly higher globe salvage rate for group D eyes in patients who received IAC compared with those who received IVC, while no statistical differences were found in groups A, B, C, and E eyes. IAC was associated with a slightly higher overall success rate than IVC. There was no difference in tumor recurrence and  IAC showed a significant advantage in globe salvage of groups IV and V eyes, but not in group I-III eyes. The overall success rate was higher in IAC than IVC. The aim of this meta-analysis was to quantitatively merge the congeneric rates of included studies that were searched as thoroughly as possible to enlarge the sample size for a more reliable result. To the best of our knowledge, this is the first meta-analysis to compare the outcomes of IVC and IAC for the treatment of retinoblastoma. We made use of multiple statistical rates to compare IVC and IAC across different staging systems to obtain an objective conclusion.
The outcomes of globe salvage of groups A to C eyes may not be reliable because of the unequal or excessively small number of trials. Overall, IAC had significant advantages over IVC in globe salvage of group D eyes and a better overall success rate. However, in approximately half of the group E eyes, treatment with IAC or IVS was unsuccessful, and the eyes eventually had to be removed. To save more diseased eyes at advanced stages, appropriate treatments should be developed. We will focus on this issue in a future study.
There was no significant difference between IAC and IVC regarding tumor recurrence. Extensive recurrence of subretinal or vitreous seeds was the most common cause of treatment failure which led to final enucleation. Shields et al. [11,30] reported that most recurrences were discovered at the three-year follow-up. In the present study, the follow-up period of patients who received IAC was shorter than three years; therefore, the effect of IAC on tumor recurrence may be over-evaluated.
Some studies have reported adverse events associated with IVC; however, most adverse events would disappear after symptomatic treatment. Rational chemotherapy drug use was essential to reduce the occurrence of adverse events in IVC treatment. In contrast, despite the advantages regarding tumor control, IAC carried a higher risk for potential local complications because of the high concentration of chemical drugs in the eye. The main temporary IACrelated complications that have been reported include eyelid edema, blepharoptosis, forehead hyperemia, and forehead hyperpigmentation, with a mean remission of two weeks to four months [28,30]. Moreover, IAC-induced vascular events, including vascular injury, spasm, obstruction, and a series of related organ ischemic lesions, deserve attention. Evidence also exists that IAC causes fatal side effects such as stroke or limb ischemia, though they are rare. Most retinal arterial obstructions were found at the one-month follow-up, whereas choroidal vascular atrophy develops slowly and usually takes several months to become apparent [40]. Vascular compromise could lead to poor vision [41], but assessment of vision requires a long follow-up. Therefore, closer visual observation should be recommended for these patients. In addition, the incidence of vascular events can be significantly reduced by precise surgery, careful angiography analysis, and ideal microcatheter placement.
In countries with advanced health care, the incidence of metastasis in children with retinoblastoma is less than 10% [10]. The risk of metastasis greatly increases with histopathologic evidence of high-risk features. Kaliki et al. [22] reported that metastasis did not occur in any patient classified with non-high-risk retinoblastoma, but that death from metastasis occurred in 4% of high-risk patients [42]. Besides, patients with heritable retinoblastoma also have an increased risk of SMNs. The most commonly observed SMNs are sarcomas, melanomas, and myelogenous leukemia, as well as cancers of the nasal cavity, orbit, and brain. The cumulative mortality ratio of an SMNs in patients with heritable retinoblastoma to those with non-hereditable retinoblastoma was 22.5:1 at 50 years of age following a diagnosis of retinoblastoma [13]. As for patients treated with IAC, accumulated exposure to irradiation and the use of melphalan may induce germline mutations and lead to SMNs [43,44]. Concerning patients treated with IVC, etoposide and carboplatin have been shown to increase the risk of SMNs [45][46][47]. Moreover, the total dose is also an important factor that should be considered when assessing risk. Most patients with SMNs received higher total doses than patients who received normal treatment. However, in all studies of SMNs, a large proportion of the eyes previously or simultaneously received external radiotherapy. Therefore, the occurrence of SMNs was not necessarily due to chemotherapy; radiation therapy may have been responsible. The effect of IAC and IVC on the prevention of SMNs warrants optimism.
This study had some limitations. Firstly, it was based on data from clinical trials which implied the impossibility of excluding the presence of confounding factors, such as the selection of medications, the severity of the condition (unilateral or bilateral disease, tumor diameter and thickness, subretinal fluid, and subretinal or vitreous tumor seeds), and previous therapies received. In addition, because retinoblastoma is a relatively rare disease, the number of eligible trials was not sufficient for a robust analysis. Moreover, the follow-up time for patients who received IAC was shorter (23.9 months) than for those who received IVC (55.6 months), which could bias the results. It was evident that long-term survival rates gradually declined as the follow-up time extended [19,25,30]. Consequently, the results of this metaanalysis were preliminary. Further analyses with more eyes and long-term dynamic observations are necessary to refine the results.