Splenomegaly (SPML) in polycythemia vera (PV): its clinical significance and its relation to symptoms, post-polycythemic myelofibrosis (PPMF) and survival

Splenomegaly (SPML) in PV is important because it has been considered a diagnostic criterion¹, yet its frequency reported in the literature ranges from 7% to 90%^1,2. Its presence influences therapy decisions^3,4 and the design of clinical trials⁵. Significant SPML in newly diagnosed PV may signal disease evolution to PPMF⁶. Therapeutically, this is relevant, because interferon may be effective in reversing fibrosis^7,8. Since review of published literature related to these issues revealed a paucity of data and variable findings we therefore studied SPML in our PV patients at the time of diagnosis (DX) or presentation (PRES). We determined its frequency and correlated this finding with other parameters of disease, including symptoms, thrombosis, development of PPMF, and survival. As a tertiary referral center, the distinction between DX and PRES is important, since it could account for some spleen size variation previously reported in the literature.

This single-center retrospective study was approved by the WCM institutional review board. We reviewed PV records in our research data repository based on an automated query system which aggregated longitudinal clinical information for analysis of spleen size according to methods previously described⁸. Standardized diagnostic criteria for PV were used for all pts⁹. Measured spleen size on physical exam was recorded in centimeters (cm) in the midclavicular line below the left costal margin. Degree of SPML was categorized into 3 subgroups: (1) “No SPML,” spleen edge not palpable or “thought” to be palpable <1 cm; (2) palpable 1-5 cm; and (3) palpable ≥5 cm. If an ultrasound had been performed, the spleen size was similarly categorized based upon the conversion method and verified formula of Chow, et al¹⁰.

Using expression matching, we extracted from our repository splenic, abdominal, and/or constitutional symptoms reported at the time of DX or PRES. A wide range of expressions were used for language-variation in symptom reporting. Symptoms had been listed in our detailed medical questionnaire for all consultation patients. They were supplemented by recorded verbal patient complaints or by physician query. Results were manually verified. Symptom severity could not be retrospectively quantified. The Myeloproliferative Neoplasm Symptom Assessment form (MPN-SAF)¹¹ was used after its publication by most but not all consultants.

Spleen size was correlated with age, sex, race, symptoms, complete blood count, JAK2^V617F allele frequency, treatment, and ELN risk score. Thrombosis-free, myelofibrosis-free (MFS), and overall survival (OS) were determined using standard Kaplan-Meier and log-rank methods. PPMF was defined by ELN/IWG-MRT criteria¹². Because we previously noted that increasing spleen size correlated with increasing marrow fibrosis¹³, we correlated spleen size at DX or PRES with marrow fibrosis grade. Multivariable analysis of spleen subgroups and the aforementioned parameters was performed using Cox Proportional-Hazards.

Our automated data query identified 517 PV pts as of March 31, 2022. 89 pts (17%) records contained no reference to spleen size at DX (n=27) or PRES (n=62) and were excluded. 25 PRES pts were also excluded because they had evidence of PPMF during the first two years of evaluation. The remaining 403 pts met our rigorous criteria for the diagnosis of PV⁹ and required phlebotomy and/or cytoreductive therapy.

Demographics are shown in Table 1. There were 176 pts (44%) in the DX subgroup and 227 (56%) in the PRES subgroup. Of the 176 DX pts, 128 (73%) had spleen size <1 cm compared to 139 (61%) of the PRES pts. SPML ≥5 cm was more common at PRES (15%) than at DX (6%). The ages of the PRES spleen subgroups (<1cm, 1-5cm, and ≥5cm) were similar at 62, 59, and 64 yrs, respectively. By contrast, the median ages of the DX subgroups showed a progressive decrease, 61, 51, and 41 yrs, respectively. A history of thrombosis in 27% of pts was similar in the 3 subgroups, as was the ELN risk status. JAK2^V617F allele frequency was significantly higher with increasing SPML (36%, 47%, and 76% respectively, p<0.001). A full molecular panel was performed on 115 patients. TET2 was significantly more common in ≥5 cm subgroup (58% vs 30% and 16%, in SPML <1 and 1-<5cm respectively, p=0.012). No other co-occurring mutations were found statistically significant, but small sample size limits this analysis.

Table 1:

Demographics by spleen size of 403 PV patients

		Overall	Spleen Size			p-value
			<1 cm	≥1 - <5 cm	≥5 cm
		n	267	92	44
DX patients 1		176 (44)	128 (73)	38 (22)	10 (6)
PRES patients 1		227 (56)	139 (61)	54 (24)	34 (15)
Age, yr 2		61 [22-91]	62 [25-91]	58 [22-85]	61 [30-85]	0.024
	DX	59 [26-91]	61 [29-91]	51 [26-77]	41 [30-82]	0.048
	PRES	61 [22-92]	62 [25-92]	59 [22-85]	64 [38-85]	0.090
Sex 1						0.7
	Female	200 (50)	136 (51)	44 (48)	20 (45)
	Male	203 (50)	131 (49)	48 (52)	24 (55)
Race/Ethnicity 1						0.3
	White	318 (79)	215 (80)	76 (83)	27 (62)
	Asian, Black, or Hispanic	38 (9)	26 (10)	7 (7)	5 (11)
	Other	1 (1)	0 (0)	1 (1)	0 (0)
	Unknown	46 (11)	26 (10)	8 (9)	12 (27)
Thrombosis history at DX or PRES 1		110 (27)	70 (26)	29 (32)	11 (25)	0.6
Time from diagnosis to PRES, yr 2		2 [0-30]	1 [0-26]	2 [0-27]	5 [1-30]	<0.001
Follow up duration, yr 2		5 [0-24]	5 [0-16]	5 [0-24]	4 [0-16]	0.2
	DX	4 [0-24]	4 [0-16]	5 [0-24]	6 [2-10]	0.4
	PRES	5 [0-16]	6 [0-16]	5 [0-16]	4 [0-16]	0.11
ELN Risk Score at DX 1						0.064
	Low	216 (54)	132 (49)	57 (62)	27 (61)
	High	187 (46)	135 (51)	35 (38)	17 (39)
JAK2 V617F VAF at DX or PRES 3		42 [27]	36 [24]	47 [28]	76 [23]	<0.001
CBC at DX or PRES 3
DX
	HCT (%)	48 [6]	49 [6]	47 [7]	47 [6]	0.3
	HGB (g/dL)	15.7 [2.0]	15.9 [1.8]	15.3 [2.3]	15.2 [2.2]	0.2
	RBC (106/uL)	5.7 [0.8]	5.8 [0.8]	5.7 [1.1]	5.6 [0.8]	0.8
	WBC (103/uL)	11 [8]	11 [8]	10 [4]	12 [8]	0.8
	PLT (103/uL)	531 [248]	531 [218]	436 [266]	558 [333]	0.3
PRES
	HCT (%)	42 [6]	42 [6]	42 [4]	43 [5]	0.14
	HGB (g/dL)	13.5 [1.8]	13.6 [1.9]	13.0 [1.4]	13.7 [1.5]	0.058
	RBC (106/uL)	5.3 [1.4]	4.9 [1.3]	6.0 [1.0]	5.8 [1.2]	<0.001
	WBC (103/uL)	13 [10]	11 [8]	21 [12]	15 [10]	<0.001
	PLT (103/uL)	484 [296]	490 [296]	468 [368]	480 [251]	0.4
Treatment 1
	IFN	77 (19)	52 (19)	17 (18)	8 (18)	>0.9
	HU	139 (34)	95 (36)	31 (34)	13 (30)	0.7
	PHL-O	149 (37)	96 (36)	34 (37)	19 (43)	0.7
	Other*	38 (9)	24 (9)	10 (11)	4 (9)	0.8

¹n (%)

²Median [range]

³Mean [standard deviation]

^*Other includes: anagrelide, imatinib, ruxolitinib and combinations

Reference to any (or no) symptoms were recorded in 374/403 (93%) pts (Supplementary Table 1). No symptoms at all were noted in 66 pts (18%), whereas 308 (82%) had 1 or more. Abdominal symptoms only were infrequent, including early satiety, nausea, left upper quadrant and splenic area discomfort occurring in 3% of pts. Conversely, constitutional symptoms without abdominal symptoms were recorded often in all subgroups. Most frequent were fatigue, pruritus, and headache (approximately 15%-20% of pts). Dizziness, night sweats, and bone pain were reported less often (5%). Both abdominal and constitutional symptoms were reported in 59 pts (16%) and more frequently in the SPML ≥5 cm subgroup (34%, p=0.002).

Marrow biopsy was performed in 212/403 patients (53%), of which 113 (53%) were at DX and 99 (47%) at PRES (Supplementary Table 2). 66 of 113 DX pts (58%) had grade 1 or 2 reticulin fibrosis, and none had grade 3. Of the 99 PRES pts, 70 (71%) had grade 1 or 2 fibrosis, and 3 (3%) had grade 3, including demonstrable collagen. Of PRES pts with SPML ≥5 cm, 80% had demonstrable reticulin and collagen was noted in 1.

Overall survival of the three subgroups was similar at 10 years (Figure 1 A,B). However, SPML at DX or PRES was associated with decreased MFS (Figure 1 C,D). The 5-yr rates of PPMF for the DX subgroups were 0%, 0%, and 10%, respectively. The 5-yr rates of PPMF for PRES was 4%, 8%, and 11%, respectively. The differences in PPMF rates between SPML groups and no SPML group were significant in multivariable analysis (1-<5cm versus <1cm: HR 2.7, p=0.01; ≥5cm versus <1cm: HR 3.2, p=0.02), independent of age, sex or treatment (Supplementary Table 3). There was no difference in OS among subgroups (multivariable analysis, Supplementary Table 4). Other multivariable and survival analyses performed as appropriate showed no association with MFS and either WBC or JAK2^V617F allele frequency, and no difference in thrombosis-free survival among spleen groups. Excluded patients with missing spleen size documentation had an MFS similar to that of the 403 included patients, suggesting they had variable degrees of SPML.

Figure 1: Overall Survival (OS) and Myelofibrosis-free Survival (MFS) by Splenomegaly Group.

A OS of DX patients, B OS of PRES patients, C MFS of DX patients, D MFS of PRES patients.

Our results indicate that splenomegaly in pts with PV is an uncommon initial finding. Approximately three-quarters of pts at DX had no SPML. Based on a small sample (n=15), clinical examination correlated satisfactorily with imaging in measuring SPML (r²=0.92). Expensive radiographic tests such as a CT appear unnecessary if the spleen is not palpable or small. However, for SPML ≥5 cm, and/or for unusual body types, radiographic studies may be required for complete evaluation. Although the numbers are small, SPML ≥5 cm was three times more common in the PRES group, perhaps related to the 2-year interval between DX and PRES. Patients who had SPML ≥5 cm at PRES had a worse MFS but not OS than those with a lesser degree of SPML or no SPML, suggesting a correlation between spleen size and MFS. This finding was independent of primary treatment received after DX or after PRES, implying SPML at DX or PRES portends a higher likelihood of progression to PPMF. This suggests that cytoreductive therapy, particularly interferon in low-risk PV pts with SPML⁴, may be justified. Accordingly, physical examination for spleen size becomes an important part of initial evaluation and follow-up of PV pts.

SPML at DX was more common in younger pts, suggesting a higher lifetime risk of disease progression to myelofibrosis. This is consistent with our previous publication indicating that myelofibrosis progression is higher than expected in younger PV pts¹⁴.

Our data show that pts with PV are highly symptomatic at onset of their illness, confirming earlier reports of others¹¹. Our symptom documentation based on retrospective chart review not quantified for intensity has obvious limitations; however, it was nevertheless very productive. It was also consistent with the subsequent MPN-SAF, which was not available during most of this study. Early symptomatic PV is another reason for modifying the widely used ELN risk score. Constitutional symptoms found at DX or PRES reflects systemic disease and relatively few symptoms specifically related to the spleen, consistent with no or limited splenic enlargement in most pts.

Finding SPML should prompt monitoring for disease progression to PPMF by monitoring symptoms, physical and peripheral blood findings. In accordance with the WHO 2016 diagnostic criteria for PV¹⁵, a marrow biopsy must be performed at DX, or at PRES if not previously performed. This fulfills not only the diagnostic criteria for PV, but also establishes a baseline for evaluating myelofibrosis, if present, and its grade. This is particularly relevant for pts at DX or PRES, with SPML ≥5 cm. MYSEC data has shown that the longer the time interval between the diagnosis of PV and that of secondary myelofibrosis, the worse the survival. Thus, careful monitoring of PV pts to detect early PPMF evolution is important since interferon has been reported a disease-modifying agent; its early use may retard or reverse PPMF⁷.

Data availability statement:

The data sets generated from our automated data repository are not publicly available since they are involved in active research. All data generated and analyzed are included in the published article.