Splenic irradiation for splenomegaly: a systematic review and meta-analysis

Abstract

Splenic irradiation (SI) is a palliative treatment option for symptomatic splenomegaly (i.e. for pain, early satiety, pancytopenia from sequestration) secondary to hematologic malignancies and disorders. The purpose of the current article is to review the literature on SI for hematologic malignancies and disorders, including: (1) patient selection and optimal technique; (2) efficacy of SI; and (3) toxicities of SI. PICOS/PRISMA methods are used to select 27 articles including 766 courses of SI for 486 patients from 1960 to 2016. The most common cancers treated included chronic lymphocytic leukemia and myeloproliferative disorders; the most common regimen was 10 Gy in 1 Gy fractions over two weeks, and 27% of patients received retreatment. A partial or complete response (for symptoms, lab abnormalities) was obtained in 85–90% of treated patients, and 30% were retreated within 6–12 months. There was no correlation between biologically equivalent dose of radiation therapy and response duration, pain relief, spleen reduction, or cytopenia improvement (r2 all <0.4); therefore, lower doses (e.g. 5 Gy in 5 fractions) may be as effective as higher doses. Grade 3–4 toxicity (typically leukopenia, infection) was noted in 22% of courses, with grade 5 toxicity in 0.7% of courses. All grade 5 toxicities were due to either thrombocytopenia with hemorrhage or leukopenia with sepsis (or a combination of both); they were sequelae of cancer and not directly caused by SI. In summary, SI is generally a safe and efficacious method for treating patients with symptomatic splenomegaly.

INTRODUCTION

Symptomatic splenomegaly is a debilitating complication commonly observed in hematologic malignancies and disorders, including chronic lymphocytic leukemia (CLL), myelofibrosis (MF), myeloid metaplasia, lymphoma, prolymphocytic leukemia (PLL), and hairy cell leukemia (HCL). In the US, splenomegaly due to these conditions represents up to 27% of all splenomegaly cases, while common benign causes include liver disease, infection, and congestive splenomegaly [1]. Manifestations of symptomatic splenomegaly include mechanical discomfort, tenderness in the left upper quadrant, early satiety, gastric compression, abdominal bloating, abdominal pain, dyspnea, fatigue, cachexia, and cytopenia secondary to sequestration [2]. First-line therapy for most patients with splenomegaly is treatment of the underlying disorder (e.g. chemotherapy or allogenic stem cell transplantation).

In spite of primary treatment, many of these patients require palliation of their splenomegaly. Palliative approaches are generally targeted at the underlying disorder (e.g. alkylating agents, immunomodulatory agents), but are often poorly tolerated and/or minimally effective [3]. Approaches directly targeting the spleen include splenectomy or splenic irradiation (SI). Although splenectomy is efficacious (and sometimes recommended for diagnosis and therapy) for select patients, 36% of those undergoing splenectomy experience significant complications, and 6% experience post-operative mortality [4].

First performed in 1903, SI is well-recognized among radiation oncologists as an alternative to splenectomy [5], though non-radiation oncologists may be unaware of its efficacy [6], [7]. SI may be an ideal treatment for patients who have a poor response to systemic therapy and/or are not surgical candidates due to advanced age or poor performance status [8]. The purpose of the current article is to review the literature SI for hematologic malignancies and disorders, including: (1) patient selection, treatment regimen, and technique; (2) efficacy of SI; and (3) toxicities of SI. Finally, we discuss the future direction of SI using novel radiotherapeutic techniques.

LITERATURE SELECTION

Evidence acquisition

We defined inclusion criteria for literature search using the Population, Intervention, Control, Outcome, Study Design (PICOS) approach [9]. We conducted a systematic search using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) literature selection process [9], [10]. We searched the medical literature from 1960 through 2016 in MEDLINE and PubMed using the terms “splenomegaly,” “radiotherapy,” and “splenic irradiation.”

After identifying studies, we excluded any (1) where the full manuscript could not be obtained; (2) not authored in English; (3) with insufficient data on outcomes or toxicities; or (4) on SI for exclusively non-malignant conditions (e.g. HIV-associated idiopathic thrombocytopenic purpura [ITP], sickle cell disease, cirrhosis). We included two studies that included patients treated for ITP (n = 4 and n = 1); the addition of these patients is relatively negligible for the entire series [11], [12]. Additionally, we included several case reports (per Table 1) in order to most comprehensively characterize treatment technique, outcomes, and toxicities.

Table 1.

Overview of disorders causing splenomegaly with associated studies and case reports.

Reference	Indication	Patients	Courses
Aabo [29]	CLL	23	31
Byhardt [36]	CLL	14	23
Guiney [15]	CLL	22	32
Nazmy [45]	CLL/CML	18	22
Roncandin [34]	CLL	38	32
Muncunill [46]	PLL	1	1
Yamamoto [47]	PLL	1	2
Al-Moundhri [48]	HCL	1	2
Dunn [49]	HCL	1	2
Nishii [50]	HCL	1	1
Sharp [51]	HCL	1	2
Weitberg [52]	HCL	1	2
Bouabdallah [30]	MPD	15	17
Elliott [32]	MPD	23	50
Greenberger [53]	MPD	14	21
Mod [54]	MPD	1	1
Parmentier [31]	MPD	9	12
Pistevoi-Gombaki [33]	MPD	20	30
Ishibashi [16]	IMF, NHL	8	8
Kriz [11]	CML, CLL, OMF, PV, AML, ITP*, NHL, MM	122	246
Lavrenkov [17]	MPD, lymphoma, CLL, HCL	32	52
McFarland [12]	CML, CLL, IMF, PV, ITPy, AML	17	26
Paulino [18]	MF, CML, CLL, AML, PV	25	25
Schratter-Sehn [19]	CLL, CML, NHL, MPD, OMF, AML	49	85
Soldic [20]	NHL, CLL, IMF	11	16
Wagner [21]	CML, IMF	17	24
Davda [35]	Waldenstrom’s macroglobulinemia	1	1

Abbreviations: AML: acute myeloid (or myelogenous) leukemia; CLL: chronic lymphocytic leukemia; CML: chronic myeloid (myelogenous) leukemia; HCL; hairy cell leukemia; IMF: idiopathic myelofibrosis; ITP: idiopathic thrombocytopenic purpura; MM: multiple myeloma; MPD: myeloproliferative disorder; NHL: non-Hodgkin’s lymphoma; OMF: osteomyelofibrosis; PLL: prolymphocytic leukemia; PV: polycythemia vera.

^*Included n = 4 patients for ITP.

^yIncluded n = 1 patient for ITP.

We further grouped the retrospective series and case reports by the decade published to provide perspectives on the evolution of SI approaches with respect to outcomes, toxicities, and multidisciplinary care. Compiling all studies in our review, a total of 486 patients were treated with 766 individual courses of SI (“course” defined as a prescribed regimen of SI with the intent to improve symptoms after therapy), with further courses (i.e. re-treatments) possible if the patient had recurrence of symptoms or a poor response. Altogether, 27 studies were included in the entire meta-analysis.

Data abstraction and analysis

With respect to indications for SI and patient setup, the following factors were obtained from the studies: patient age, indication for SI, prior treatment (e.g. chemotherapy), percent of patients presenting with splenic pain or CBC irregularities, SI beam arrangement median dose per fraction and course, total courses received per patient, and toxicities according to Radiation Therapy Oncology Group (RTOG) toxicity grade. Individual patient data were not available for all patients (particularly those taken from larger series).

With respect to doses, we present radiation therapy (RT) regimens as reported in the studies, including patient setups, dose per fraction, dose per RT course (with the median dose used, if a range was provided). The α/β ratio is used in the calculation of the biologically equivalent dose (BED), which helps to compare different fractionation schedules:

In this equation, n is the number of radiation fractions and d is the fraction size. An α/β ratio is used to estimate the effects of radiation on various tissues and compare various dose and fractionation schemes [13], [14]. We used the α/β ratio value of 10 (i.e. BED10) in our calculation because this approximates responses for cancer and acute toxicities.

With respect to outcomes, we abstracted the rate of complete response (CR) or partial response (PR) of the symptom for which SI was prescribed (e.g. splenomegaly, pain, cytopenias). Additionally, the duration of the response – defined as clinical pain relief or reduction of splenomegaly – was provided. With respect to toxicity, the rate of any RTOG grade 1–5 toxicity was calculated based on each study. The underlying disorders of these patients, splenomegaly, and SI may all cause leukopenia and anemia; unfortunately, we were unable to discern the etiology of the various reported cytopenias. We reported toxicities per definitions of the Radiation Therapy Oncology Group (RTOG), as per Supplementary Table 1. Notably, certain toxicities per the RTOG scale (e.g. sepsis secondary to leukopenia) may be from the treatment (e.g. SI, prior chemotherapy) or from disease progression. In some cases, cytopenias were considered to be a result of progressive disease rather than SI toxicity, and this was not coded as toxicity. Finally, we identified predictors of poor response across the various disorders.

PATIENT AND TREATMENT FACTORS

Table 1 summarizes the studies included in this review, including the indication for SI. There were 766 SI courses to 486 patients, most of whom were between 50 and 70 years old (median age = 60.5 years). One study alone reported 122 patients and 246 courses [11]. Symptomatic splenomegaly was the indication for treatment among nearly all studies. Two studies include one patient each who received SI in the absence of splenomegaly [15], [12]. Among all studies included, patients were most commonly treated for the following diseases: CLL/CML (n = 115; 23.7%); PLL (n = 2; 0.4%); HCL (n = 5; 1.0%); MPD (n = 82; 16.9%); additionally, some studies treated patients for various etiologies, including those already listed such as polycythemia vera, multiple myeloma, idiopathic thrombocytopenic purpura, and non-Hodgkin’s lymphoma (n = 281, 57.8%) [11], [12], [16], [17], [18], [19], [20], [21]. All treatment courses were palliative by intent. The most common indications were pain or discomfort (n = 469; 97%) from splenomegaly and cytopenias (e.g., anemia, thrombocytopenia, leukopenia, or lymphocytosis; n = 298; 61%).

Overall, 58% (n = 280) of patients received treatment for splenomegaly prior to RT. Commonly cited therapies were prednisone (n = 56), hydroxyurea (n = 48), chlorambucil (n = 29), busulfan (n = 18), and cyclophosphamide (n = 13). Systemic therapy was administered concurrently with RT to 4 patients in one study [21]; RT was used alone in all other studies.

Data on number of courses delivered was available for 64% (309/486) of all patients included in the study representing 57% (435/766) of all courses given. There was variability in both the dose per fraction (range: 0.1–2.5 Gy) and the total dose (range: 0.15–30.5 Gy). The most common fractionation regimen was 1 Gy per fraction, to a total of 10 Gy, which corresponds to a BED10 of 11 Gy. In juxtaposition, the BED10s of external beam RT for prostate cancer are typically 80–90 Gy [22], and >120 Gy with brachytherapy [14], [23], [24], [25]. The BED10 of stereotactic body RT for lung cancer is ∼100 Gy [26], [27], [28]. Other common fractionation regimens for SI used even lower doses, often 0.5 Gy per fraction, totaling to about 6 Gy (BED10 of 6.3 Gy) per course with retreatment as needed for symptomatic control. Fractionation was most often daily or every other day (i.e., 3 times/week).

Of 309 patients who could be analyzed for number of courses received, 27% (n = 83) received multiple courses of RT (i.e. either because of no response with the first course, or recurrence of symptoms); the median number of courses was 1, range 1–9. An average RT dose per course was around 9–10 Gy, yet some patients received in excess of 24 Gy by the conclusion of all treatments [29], [30], [15], [16], [31]. In general these studies cited symptoms refractory to therapy as the reason for higher doses and multiple courses.

Details about linear accelerators and patient setup were available for 42% (324/766) of all courses delivered: 77% (n = 250) were delivered by linear accelerator photon units, 16% (n = 51) by Cobalt-60 teletherapy units, and 7% (n = 23) with electrons. Of 534 courses with field setup information, 93% (n = 497) were delivered via AP/PA fields and 6% (n = 30) via single anterior field.

OUTCOMES

Efficacy of SI was assessed based on pain palliation, splenic size reduction, and/or improvement in cytopenias (documented as a CR or PR). For studies that listed separate response rates for splenomegaly response, reduced pain, and improved cytopenias, a weighted average was calculated using the number of courses for each result. Additionally, improvement in any one or a combination of the listed factors was considered a PR with respect to the overall efficacy of SI.

Overall, 72% of courses reduced the size of the spleen, 59% of all courses resulted in splenic pain relief, and 78% resulted in improvement of cytopenias. Additionally, the median PR rate for relief of any symptom among all studies was 88%, while the mean PR rate was 85%. Fig. 1 displays partial responses in individual studies. For studies of RT for multiple indications and for CLL/CML, PR among all studies was greater than 60%. For both PLL and HCL, partial responses were 100% but these studies included significantly fewer patients. The most varied results occurred in studies on myeloproliferative disease, where PR ranged from less than 40% to 100%. Studies lacked sufficient data to determine which aspects of cytopenias responded best to SI. Studies also offered data on duration of therapeutic benefit and overall survival, and these results are included in Fig. 2. On average, common symptom relief duration was 6–12 months [30], [32], [33], [34]. Several studies reported long-term relief of symptoms, including 18 years in one study [35].

Figure 1. Partial response rates of SI in different diseases.

Legend: Partial response rates of SI in different cancers. Legend: The rate of partial or complete response from each study is plotted vs. indicated disease (e.g. CLL/CML; PLL, HCL, MPD; or studies that included multiple cancers). The size of the circle corresponds to the number of patients included (inset). There was no apparent outlier hematologic malignancy that markedly different response rates to SI vs. other malignancies. The response is determined by directly reporting from paper abstracts or calculating a response rate based on data provided in the paper. For papers that reported separate outcomes for splenomegaly, pain, and CBC responses, a weighted average was calculated based on the number of courses.

Figure 2. Efficacy and toxicity vs. BED.

Legend: Efficacy and toxicity vs. BED for all cancers. Legend: The efficacy and toxicity are plotted vs. dose of radiation among all studies. All cancers from Fig. 1 are included. The dose of radiation is calculated by using BED = (nd[1 + d/(a/b)]), n is the number of radiation fractions and d is the fraction size, and a/b is 10. (A) The response rate (with PR or CR) for any symptom alleviation vs. BED is plotted; there is no improvement in symptoms with increasing dose. The most common fractionation regimen was 1 Gy per fraction, to a total of 10 Gy, which corresponds to a BED10 of 11 Gy (vertical line coplotted). (B) The response rate (with PR or CR) for individual symptom alleviation vs. BED is plotted. With respect to outcomes, 65–100% of patients experienced pain relief, 60–100% had a reduction in spleen size, while the number with CBC improvement was variable. The median response duration was 11 months (range 2–216). There is no improvement in symptoms with increasing dose (r2 = 0.01 0.05). The size of the circle corresponds to the number of patients included (inset). (C) The toxicity rate (grade 3–5) vs. BED is plotted. The rate of any toxicity was between 5 and 60% among most studies. There is no association in symptoms with increasing dose (r2 = 0.02 0.34). The size of the circle corresponds to the number of patients included (inset).

Supplementary Table 2 lists factors associated with poor response, including high transfusion requirements (e.g. >2–4 units per month) [30], prior use of chemotherapy [36], [15], diagnosis of high-grade lymphoma [17], treatment to <5 Gy [18], [30]. Supplementary Table 3 presents a comprehensive review of outcomes and toxicities. Per Fig. 2, there was no apparent correlation with improvement in symptoms of doses of 5 Gy vs. 10 Gy, delivered in 1 Gy fractions. Patients who previously failed chemotherapy also had worse response to SI retreatment [36], [15]. Age and RT field size were not associated with response [20].

Fig. 2 shows plots of outcomes and toxicities vs. BED10. There was no relationship between BED10 response duration or the proportion of courses in each study displaying pain relief, reduced spleen size, or CBC improvement. Similarly, there was no clear relationship between BED10 and toxicity (i.e., all r2 values <0.4). There was no apparent outlier in poor outcomes based on year of study publication. The most common regimen was 10 Gy in 1 Gy fractions over two weeks, and the data suggest that lower doses (e.g. 5 Gy in 5 fractions) may be as effective as higher doses. This low dose of RT is similar to that used the treatment of certain benign diseases (e.g. sialorrhea) [37]. These findings are consistent with the plateau of the sigmoidal dose-response curve seen in other disease sites where RT results in excellent local control, including T1 lung cancer [38], metastatic disease treated with stereotactic radiation [26], [27], and prostate cancer [14], [22], [39].

TOXICITIES

Table 2 lists data on SI toxicity according to RTOG criteria. Only acute toxicities were calculated; no studies reported long-term toxicity. The most common toxicities were grade 3 and 4, constituting 82.5% (n = 142/172) of all reported toxicities. However, when compared to the total sample population, these effects account for adverse effects in only 18.5% (n = 142/766) of all SI courses. The most common toxicities were hematologic (n = 162, 94% of documented toxicities), including isolated neutropenia (n = 6, 3%), anemia (n = 48, 28%), thrombocytopenia (n = 52, 30%), or leukopenia (n = 36, 21%); a combination of leukopenia and thrombocytopenia (n = 7, 4%); and pancytopenia (n = 13, 8%). One course resulted in non-lethal myocardial infarction resulting from radiation induced anemia [12]. Notably, there were 5 cases (0.7% of all courses) that resulted in death (grade 5 toxicity per RTOG criteria); all were due to either (1) thrombocytopenia with hemorrhage [12], (2) leukopenia with sepsis [15], or (3) a combination of both [32]. Thus, the grade 5 toxicities were sequelae of cancer and not directly caused by SI. The other documented toxicity was nausea which was either self-limited or treated with anti-emetics. There was no apparent different in toxicities after retreatment vs. the first session of SI.

Table 2.

RTOG grade toxicities after SI.

Symptom	n	Grade 1 or 2	Grade 3 or 4	Grade 5	Low Grade % (1 or 2) / High Grade % (3–5)
Nausea/vomiting/abdominal cramps	9	8	1	0	89% / 11%
Leukopenia/neutropenia and/or thrombocytopenia	114	5	104	5	4% / 96%
Anemia	48	12	36	0	25% / 75%
Cardiovascular (myocardial infarction)	1	0	1	0	0% / 100%
Totals	172	25	142	5	15% / 85%
Toxicity-free courses	N = 594 (77%)

The bold values represent the totals.

FUTURE DIRECTIONS

Our study findings are consistent with a published 2004 poll of radiation oncologists in the United Kingdom, in which parallel anterior-posterior fields were the preferred technique (54.3%) and photons were the most frequently used source (94.7%), using 10 Gy in 1 Gy fractions [40]. Commonly reported toxicities were asymptomatic pancytopenia, GI upset, and anemia requiring transfusion, which are findings consistent with our review. Finally, this study discussed a few options providers employ to regularly monitor blood counts, with a majority of the polled oncologists (89.5%) regularly monitoring complete blood counts and discontinuing therapy if a prescribed cutoff is reached.

Following RT of relatively small doses (i.e. <2 Gy), cancers of the hematologic system may respond dramatically due to intracellular death pathways not typically found in carcinomas (e.g. apoptosis, necroptosis) [41]. The treated organ (i.e. the spleen) may decrease in volume dramatically between fractions. Thus, over the coming decades, the use of adaptive RT (i.e. the reduction of RT field size prior to each fraction) may play a larger role in SI. Field size reductions may reduce toxicity while being equally effective as non-reduced fields. Several of the studies included in this review reduced field size during the course of treatment, though details were limited [29], [19], [30], [20]. Future studies may explore this technique more using adaptive planning with inter-fraction assessment (Supplementary Fig. 3C).

LIMITATIONS

There are several limitations to our analysis. First, we included a heterogeneous group of patients, with heterogeneous indications for treatment (e.g. pain, cytopenias). Additionally, two studies included patients who were treated for ITP, a non-malignant cause of splenomegaly [11], [12]. Moreover, many of the papers included, particularly the larger studies, did not include patient-level data for analysis. Finally, the toxic effects of SI are similar to the sequelae of hematologic malignancies. In our analysis of toxicity following SI, it was usually not possible to discern which patients experienced RTOG-grade toxicities due to progressive disease versus RT, and certain deaths may be due to the cancer itself rather than non-cancer causes (e.g. low hemoglobin causing myocardial infarction) [42]. Further, the BED equation may not adequately characterize extremely hyperfractionated or hypofractionated regimens (e.g. >8 Gy/fraction), cellular death due to different modes surrounding mitotic catastrophe (e.g. necroptosis) [41], effects on stroma/vasculature (e.g. pericytes) [43], or molecular pathways behind recurrence (e.g. vasculogenesis) [44]. Additionally, the BED does not take into account the volume of irradiated tissue or how the dose is prescribed (e.g. to a volume vs. isodose line).

This review also reveals that while effective, certain patients SI can have significant toxicity. Our analysis of toxicity reveals that grade 3 and 4 toxicities are more commonly reported than low-grade toxicities, although this may be due to underreporting. Supplementary Fig. 3 offers a reasonable treatment schema that is consistent with the results of our literature survey. A potential option would be to deliver 5 Gy in 5 fractions. Given the limited hematologic reserve of these patients, clinicians should follow these patients closely and have a low threshold for treating the patients for infection, thrombocytopenia, and anemia.

CONCLUSION

SI is a safe and efficacious method for treating symptomatic splenomegaly from hematologic malignancies. The most common dose regimen is 10 Gy in 1 Gy fractions over two weeks. Lower doses (e.g. 5 Gy in 5 fractions) may be equally efficacious. Given the limited hematologic reserve of these patients, clinicians should follow these patients closely and have a low threshold for treating the patients for infection, thrombocytopenia, and anemia.