P. Fenaux, Service d'hématologie clinique, Hopital Avicenne/Paris 13 University, 125 rue de Stalingzad 93009 Bobigny, France. E-mail: firstname.lastname@example.org
Summary. Splenectomy remains the most effective treatment of chronic autoimmune idiopathic thrombocytopenia (ITP) (i.e. of > 6 months duration). Treatment of patients refractory to splenectomy (with absence of response or relapse after initial response) is difficult, and their long-term outcome is not well known. Over a 10-year period, 183 patients with chronic ITP were splenectomized including 158 adults and 25 children (≤ 15 years). Forty-seven of them, who were refractory to this treatment, were prospectively followed up for 5--15 years (median 7·5 years). Twelve of them, with moderate thrombocytopenia, remained untreated, and 35 were treated by a median of two regimens (range 1--6), to which 27 responded. Thirty-six (77%) of the refractory cases reached platelet counts durably > 100 × 109/l, nine of them without treatment and 27 of them with low-dose steroids or azathioprine; six (13%) remained moderately thrombocytopenic (35 × 109/l to 100 × 109/l platelets); the last five patients, without response to any treatment (up to six regimens), remained severely thrombocytopenic (platelets < 20 × 109/l), and three of them died from bleeding. Twenty-seven (57%) of the 47 refractory cases required at least one hospitalization, in the majority of cases for intravenous immunoglobulin (IVIg) infusions. Seven of the refractory cases occurred in children. Six of them subsequently reached platelet counts > 100 × 109/l, but one died from bleeding. Our findings confirm the overall favourable long-term prognosis of chronic ITP refractory to splenectomy.
Autoimmune idiopathic thrombocytopenia (ITP) is characterized by specific platelet autoantibodies (Godeau & Bierling, 1996; Karpatkin, 1997), leading to peripheral destruction of platelets, which are mainly splenic. ITP can have a favourable evolution within 6 months of diagnosis (acute ITP, occurring mainly in children) (Godeau & Bierling, 1996). The persistence of thrombocytopenia for longer than 6 months characterizes chronic ITP, which is mainly seen in adults.
Splenectomy remains the best treatment of chronic ITP, but approximately 30% of patients do not respond or relapse after surgery (Fenaux et al, 1989; Stasi et al, 1995; Karpatkin, 1997; George et al, 2000), and are defined as refractory to splenectomy. Treatment remains difficult in such patients. Some of them respond to steroids or immunosuppressive drugs (cyclophosphamide, cyclosporine, azathioprine), androgens, vinca alkaloids or high-dose intravenous immunoglobulins (IVIg), but long-term results of such treatments in a large series of chronic ITP refractory to splenectomy are not well known. We studied prospectively the long-term outcome of patients refractory to splenectomy in a cohort of 183 consecutive cases of chronic ITP that were splenectomized over a period of 10 years.
Patients and methods
Between 1 January 1985 and 31 December 1994, the diagnosis of chronic ITP (i.e. of > 6 months duration) was confirmed at our institution by a platelet lifespan study in 255 unselected patients with suspected ITP diagnosis. Measurement of platelet lifespan and study of the sequestration site of 111In-labelled platelets were performed as described previously (Fenaux et al, 1989).
Seventy-two patients were not splenectomized because of platelet counts > 50 × 109/l (54 patients), contraindication to surgery (10 patients) or patient refusal (eight patients).
One hundred and eighty-three patients were splenectomized. In all, 65% of patients were female and 35% males. Median age was 32 years (range 4--75 years). A total of 158 adults underwent splenectomy and 25 children (≤ 15 years), of which 14 were female and 11 male.
All patients received a pneumococcal vaccine before surgery and received post-operative daily penicillin.
Response to splenectomy was defined by platelet counts > 100 × 109/l 3 months after surgery, and relapse by platelet counts < 100 × 109/l in initial responders. Refractoriness to splenectomy was defined by an absence of initial response or subsequent relapse after splenectomy.
Treatment strategy, in patients refractory to splenectomy, depended on platelet counts and bleeding symptoms. Patients with platelet counts above 35--50 × 109/l and with no bleeding received no treatment. Patients with severe bleeding received short-term IVIg courses (0·4--1 g/kg for 1 or more days). Patients with platelets below 35--50 × 109/l or with moderate bleeding were treated, generally with steroids, and in case of failure, successively with azathioprine, vinca alkaloids, danazol and, less often, drug combinations or investigational drugs.
Response to splenectomy
One hundred and fifty-nine patients (87%) had response to splenectomy at 3 months, and 24 (13%) patients had initial failure. Four of them subsequently achieved normal platelet counts without treatment. Neither post-operative death nor subsequent fatal infections were seen.
Twenty-nine (16%) of the responders subsequently relapsed, 86% of them within 3 years of surgery. Thus, 53 (29%) patients were considered refractory to splenectomy.
Analysis of prognostic factors of splenectomy, including age at the time of diagnosis, sex, platelet counts before splenectomy, platelet counts 10 d after surgery, platelet lifespan and site of platelet sequestration, was made using the chi-square and Student's tests (Table I). The only significant prognostic factors of refractoriness to splenectomy were a platelet count below 50 × 109/l 10 d after surgery (P < 0·001) and liver predominance of platelet sequestration (P = 0·025).
Table I. Prognostic factors of initial response to splenectomy in 183 patients with chronic ITP.
Age (years, mean)
Platelet count (mean, × 109/l)
10 d after splenectomy
(mean no. of days)
Predominant sequestration site of labelled platelets
(mean spleen/liver ratio)
Seven children (four female and three male) failed splenectomy, including two initial failures and five relapses.
Evolution of the cases refractory to splenectomy
Six of the refractory patients were lost to follow-up between 3 and 17 months (median 8 months).
The remaining 47 patients were followed up between 5 and 15 years (median 7·5 years) (Table II).
Table II. Initial characteristics and outcome of the 47 cases of chronic ITP refractory to splenectomy.
Interval from diagnosis to splenectomy (months)
Initial failure (IF) or time from splenectomy to relapse (months)
Untreated patients. Twelve patients never received treatment. Nine of them eventually reached durable platelet counts > 100 × 109/l. The last three patients remained thrombocytopenic (37--80 × 109/l). Only one patient had one moderate bleeding episode (nosebleed), when the platelet count was 30 × 109/l: there was no treatment because of patient refusal.
Treated patients. The other 35 refractory cases were treated with different approaches, including steroids, azathioprine, danazol, continuous IV vinblastine, interferon, cyclosporine, high-dose steroids and removal of an accessory spleen (Simon et al, 1987; Fenaux et al, 1990; Facon et al, 1991; Caulier et al, 1995). Results of those treatments are shown in Table III. Eighteen patients, including seven of the 32 patients treated with steroids (prednisone, 1--2 mg/kg daily for 3 weeks followed by slow dose tapering), five of the 17 patients treated with azathioprine (oral daily dose of 2.5 mg/kg, over several months), one of the nine patients treated with danazol (600 mg/d), two of the three patients with accessory splenectomy, one of the six patients treated with continuous vinblastine IV infusions (0.1 mg/kg weekly), one of the three patients treated with an association of azathioprine and danazol, none of the two patients treated with alpha interferon (3 million units 3 d a week for 4 weeks), none of the three patients treated with high-dose IV methylprednisolone (1 g/d, for 2 d) and the patient treated with oral cyclosporine, all achieved a durable response, i.e. persisted after drug discontinuation. Nine additional patients were also durable responders but required continuous drug administration of low-dose steroids (five patients), azathioprine (two patients), danazol (one patient) and danazol plus azathioprine (one patient). Twenty-two patients received at least one course of IVIg with a transient response occurring in 15 (68%) of them.
Table III. Results of therapeutic approaches in patients refractory to splenectomy (present study and literature).
Thirty-six (76·5%) of the 47 refractory patients eventually reached platelet counts > 100 × 109/l, six (12·7%) had moderate and stable thrombocytopenia with platelet counts between 35 × 109/l and 100 × 109/l. The remaining five (10·6%) patients (truly refractory cases) remained severely thrombocytopenic (< 20 × 109/l). After splenectomy, they received 0--6 treatments (median 3), to which four never responded, whereas one had a transient response to steroids.
Twenty-seven of the 47 patients (57%) were hospitalized at least once for their ITP. None of the untreated patients required hospitalization. In patients who required treatment, the median number of hospital stays was five (range 1--35). The five ‘truly refractory’ cases were hospitalized between one and 54 times (median 5).
In most cases, hospitalizations were motivated by a need for IVIg infusion because of severe thrombocytopenia without severe bleeding. Only three patients were hospitalized for severe bleeding. Three patients (6% of the refractory cases and 1·6% of the splenectomized patients) died from bleeding 13, 53 and 75 months after splenectomy. The first patient, aged 10 years at diagnosis, presented with two late relapses: the first one 6 years after splenectomy, which responded to steroids, and the second one, 7 months later, where he presented with spontaneous and sudden central nervous system (CNS) bleeding with no known predisposing factor, which was fatal within a few hours. The second patient, aged 73 years, presented with an early relapse, 5 months after splenectomy. He was treated with azathioprine without response and died a few months later of CNS bleeding. The third patient, aged 51 years, presented with an initial failure of splenectomy. Many treatments were attempted including steroids, danazol, vinblastine, combination of danazol and azathioprine, but with no efficacy. This patient, known to have a history of gastric ulcer, died from massive gastrointestinal bleeding.
Specific outcome of children refractory to splenectomy
Seven of the 25 splenectomized children (three males and four females) failed splenectomy, including two initial failures and five relapses. Six of these seven children subsequently achieved durable platelet counts > 100 × 109/l after steroids alone (four cases), steroids and azathioprine (one case) and multiple successive treatments including steroids, azathioprine, danazol, vinblastine (one case). As reported above, the last child, aged 10 years, presented with two late relapses and died from fatal CNS bleeding. Five of the children received IVIg. Six of them were hospitalized at least once for their ITP (the total number of hospitalizations in the six children was 12). Hospitalizations were motivated by the need for IVIg infusions (6) or bleeding (6).
Our results confirm the 70--80% response rates after splenectomy that have been reported in previous studies (Stasi et al, 1995; Godeau & Bierling, 1996; McMillan, 1997; Mazzucconi et al, 1999; Fabris et al, 2001). The prognostic factors of refractoriness to splenectomy, i.e. platelet counts below 50 × 109/l 10 d after surgery and liver predominance of platelet sequestration, had also been found in most previous studies (Fenaux et al, 1989; Najean et al, 1991; Lamy et al, 1993; Radaelli et al, 2000). In our series, the proportion of children cured by splenectomy was similar to that observed in a review of 16 series involving 271 children undergoing elective splenectomy (72%) (George et al, 1996). Neither post-operative death nor subsequent fatal infections were seen in this population. This is somewhat different from a recent report on morbidity and mortality in adult ITP (Portielje et al, 2001), in which 26% of early post-operative complications (including pulmonary embolism, intra-abdominal bleeding, Gram-negative sepsis, abdominal wall haematoma or abscess, pneumonia and pneumococcal sepsis) were seen, and where the incidence of post-operative death was 2·5%. Pizzuto & Ambriz (1984) reported a 7·8% complication rate of splenectomy including subphrenic abscess, thromboembolism and visceral lesions. However, they observed only one death (0·25%) caused by fatal sepsis. In most other series, post-splenectomy complications were also generally rare (den Ottolander et al, 1984; Stasi et al, 1995), as in our study.
When treatment is required in patients who are refractory to splenectomy, the choices are uncertain. Many studies have reported results of different therapeutic approaches including steroids, immunosuppressive drugs, androgens, vinca alkaloids, high-dose immunoglobulins (Godeau et al, 2002) and, recently, even anti-CD20 monoclonal antibodies (Stasi et al, 2001). Their results are summarized in Table III and are generally similar to those observed in our patients who were refractory to splenectomy. However, in most papers, the results of a single treatment were reported, the number of patients were rather small, patients had previously received a great variety of treatments, and follow-up was variable.
Only a few studies in ITP have analysed the long-term outcome of a patient population that underwent successive treatment and included follow-up periods. In addition, most of those studies focused on a population first analysed at the time of diagnosis of ITP, i.e. when only a minority of cases were still thrombocytopenic after splenectomy (Table IV).
Table IV. Long-term outcome of patients refractory to splenectomy (present series and literature).
We prospectively analysed a relatively large population of ITP patients that were refractory to splenectomy, i.e. the chronic ITP population with a potential bleeding risk. In those patients, we used a treatment strategy based on our previous experience and on guidelines suggested by different authors, including those of George et al (1996). No treatment was given to patients with platelet counts above 35 × 109/l to 50 × 109/l and without bleeding. Patients with lower platelet counts (or an additional cause of bleeding) were generally treated with the following steps: steroids followed, in case of failure, by azathioprine, danazol, vinblastine infusions, combinations of those drugs or investigational drugs. At some point, the presence of an accessory spleen was often assessed. IVIg was restricted to patients with bleeding or severe thrombocytopenia.
Using this strategy, our major finding was that, with a minimum follow-up of 5 years (median 7·5 years), 42 of the patients refractory to splenectomy reached a platelet count durably > 35 × 109/l (36 of them had platelet counts > 100 × 109/l). Six of the seven refractory childhood cases achieved platelet counts durably > 100 × 109/l.
At the time of last follow-up, 33 of these 42 patients had ceased treatment, whereas nine required continuous drug administration, which was mainly low-dose steroids or azathioprine. Nineteen of the 42 patients required at least one hospitalization, but these were mainly for IVIg infusions. Thus, only five (10%) of the 47 refractory cases, representing 2·6% of the splenectomized cases, had severe thrombocytopenia at the end of follow-up. These patients required frequent hospitalizations, and three of them (1·6% of the splenectomized cases) died from bleeding.
In previously published series (den Ottolander et al, 1984; Pizzuto & Ambriz, 1984; Stasi et al, 1995; Mazzucconi et al, 1999; Portielje et al, 2001), the proportion of patients refractory to splenectomy who, at the end of follow-up, had platelet counts < 100 × 109/l ranged from 59% to 76%, i.e. more than the 39% observed in our series. To our knowledge, only one of these studies focused precisely on the follow-up of splenectomized patients (Mazzucconi et al, 1999). The authors noted bleeding symptoms in 15 of the 24 patients who relapsed, which was controlled by very low doses of prednisone. Only limited data were available on hospitalization and quality of life. In contrast, Portielje et al (2001) observed an important incidence of ITP-related admissions (183 per 1000 person--years), especially in patients refractory to all treatments, and medical or surgical treatment of symptomatic thrombocytopenia was the cause of 80% of these admissions.
In previous reports, the overall mortality of splenectomized patients has ranged from 0·2% to 4·2%, and was 1·5% in our experience. In a recent report (Portielje et al, 2001), it was found that most adult patients with ITP had a good outcome with no excess mortality; however, patients with persistent low platelet counts below 30 × 109/l had an increased mortality risk, compared with both the general population and patients with platelet counts above 30 × 109/l. This is in agreement with a previously published report (Cohen et al, 2000).
The three deaths we observed in our study occurred in patients who did not achieve platelet counts > 10 × 109/l in spite of various treatments.
Finally, in our study, with the exception of the few patients refractory to all treatments, quality of life was generally good, and hospitalizations were both infrequent and short.
In conclusion, in spite of an absence of treatments with both major and regular efficacy in chronic ITP refractory to splenectomy, most patients have a favourable outcome, although they often maintain low platelet counts.