British Journal of Haematology, 1990. 76, 250-255
Lack of useful clinical predictors of response to splenectomy in patients with chronic idiopathic thrombocytopenic purpura A. J U L I A , ’ c. ARAGUAS,’ J. ROSSELLO,’ J. B U E N O , ‘ P. D O M E N E C H , 2 M. O L O N A , ’ R. G U A R D I A , 3 J. P E T I T 4 AND A. FLORESS ‘Hospital ‘Val1d’Hebron’, Barcelona, 2Hospital ‘Princeps d’Espanya’, Barcelona, Hospital ‘Alvarez de Castro’, Gerona, * ‘Mutua de Tarrasa’, Barcelona, Hospital ‘Germans Trias’, Badalona, Spain
Received 20 December 7 989; accepted for publication 30 May 1990
Summary. We set out to identify clinical or analytical variables that might predict the response to splenectomy in patients with chronic idiopathic thrombocytopenic purpura (ITP). We retrospectively examined 138 mostly adult patients with chronic ITP, treated with splenectomy. Information was compiled from five Public Health Hospitals from a questionnaire and chart review. Sixty-one potentially prognostic variables were analysed by univariate and multivariate statistical methods. After multivariate analysis, age (relative risk=1.02; CI 1-1.03) and a severity of the bleeding
diathesis (relative risk= 1.6; CI 1.1 3-2.22) were independent prognostic factors for a sustained response to splenectomy. An independent analysis of the postsplenectomy counts showed that an early (days 1-3) thrombocyte count exceeding 156 x 10y/lcells increases the likelihood of a permanent unmaintained response. Our data indicate that the response to splenectomy in patients with chronic ITP cannot be adequately predicted on the basis of pre-splenectomy clinical or analytical variables.
Chronic idiopathic thrombocytopenic purpura (ITP) is a disorder in which platelets are destroyed by macrophages mostly in the spleen. In contrast to the acute disorder, chronic ITP rarely resolves spontaneously and requires the choice of one of several therapeutic options. Splenectomy removes the major site of platelet destruction (Gernsheimer et al, 1989) and with corticosteroids and immunosuppressive agents constitutes the group of the three most effective forms of treatment (Pizzuto & Ambriz, 1984). Splenectomy is usually indicated if a trial of steroids has failed or if the dose requirements are such that the patient is developing serious side effects. While a general consensus exists on the indications for splenectomy (Karpatkin, 1980: McVerry, 1985),it is unfortunate that the possible benefits of the operation seem to be totally unpredictable. Several groups have tried to detect useful predictors of the effect of the operation. While many workers agree upon the unpredictability of the response to splenectomy, others have reported on variables that might help in the prognosis of the disease after the operation. Those factors more frequently mentioned as predictive ones have been: age (the younger the better)
(Aster, 1972a: Di Fino et al, 1980; Akwari et al, 1987: Fenaux ct al, 1989; Fabris et al, 1989), previous response of the disease to prednisone (splenectomy more likely to be effective) (Thompson et aI, 1972: Jiji et nl, 1973: Brennan et al, 1975; Lacey & Penner, 1977; Mintz et al, 1981; Kernoff & Malan, 1983), short diagnosis-to-splenectomy interval (shorter intervals more likely to be followed by good surgical results), shortened platelet half-lives (shorter survivals predictive of better responses) (Siegel et al, 1989) and predominantly splenic sequestration (predictive of a good response) (Najean & Ardaillou, 1971; Gugliotta et al, 1981; Goudemand e t al, 1982; Fenaux et al, 1989; Gernsheimer et al, 1989). In contrast with these reports, the following references are from studies that failed to confirm the predictive value of the same five types of prognostic factors: age (Gugliotta et al, 1981: Goudemand et al, 1982), prednisone sensitivity (Doan et al, 1960; Gugliotta et al, 1981; Goudemand et al, 1982; Fenaux et al, 1989). time to splenectomy (Gugliotta et aI, 1981; Goudemand et al, 1982; Pizzutto & Ambriz, 1984; Fenaux et al, 1989), platelet survival (Gugliottaetal, 1981; Akwarietal, 1987;Fenauxetal, 1989)and site of sequestration (Aster, 1972b; Siegel et al, 1989). The aim of this study was to see whether a comprehensive analysis of the pre-splenectomy clinical and analytical features might help to identify those patients more likely to
Correspondence: Dr A. Julia, Servicio de Hematologia. Hospital General ‘Vall d’Hebron’,Po Vall d’Hebron, sin 08035. Barcelona. Spain.
2 50
Splniectomy in Patients with Chronic ITP respond. This report describes the treatment for 138 splenectomized patients and the statistical analysis of 61 potential prognostic variables. MATERIALS A N D METHODS This was a retrospective multicentre study with patients seen at five participating institutions. Details of each patient’s history and laboratory data were retrieved from hospital records and registered on a previously designed sheet. The diagnosis of ITP was established according to criteria generally accepted: thrombocytopenia with abundant or adequate megakaryocytes in bone marrow aspirate and absence of systemic disease, splenomegaly, coagulation disorder or any other likely cause of a low platelet count (Karpatkin, 1980: McMillan. 198 1). We evaluated a total of 6 1pre-splenectomy potentially prognostic clinical and analytical variables. The clinical variables included age, sex, year of diagnosis and splenectomy. time with haemorrhagic manifestations, interval between diagnosis and splenectomy and presence of coexisting diseases not aetiologically related to autoimmune thrombocytopenia. The characteristics of the bleeding diathesis were carefully recorded. In addition to noting the presence of petechiae, bruising, epistaxis, gingival or conjunctival bleeding and other sites of bleeding, for the purposes of this investigation we graded the two more common types of bleeding into four groups. Petechiae were grouped into none, less than 10, 10-100 or more than 100 lesions. Likewise, bruising was stratified into none, less than five haematomas, five to 10 lesions and more than 10. We also grouped the haemorrhagic manifestations into three categories: skin alone, mucosal alone or skin and mucosal. The overall intensity of the bleeding was considered to be absent, mild. moderate or severe, according to a generated score: mild bleeding was defined as the presence of skin manifestations alone (few petechiae, mild bruising); moderate bleeding as the association of more intense skin lesions with one mucosal bleeding site, and severe bleeding, as the combination of severe skin manifestations plus more than one site of mucosal haemorrhage. We analysed the following laboratory data: haematocrit, haemoglobin. platelet count, white blood cell count, absolute neutrophil count, lymphocyte, monocyte and eosinophil counts, at diagnosis, 3 months and 1 month prior to and immediately before splenectomy. A variable called ‘lowest’ (lowest platelet count of all pre-splenectomy determinations) was also independently evaluated. Post-operative peripheral blood counts were also recorded to assess the response at several time intervals, until the last known follow-up visit. Other pre-splenectomy variables included bone marrow specimen analysis, and presence and type of antiplatelet antibodies. Platelet antibodies were determined by an immunofluorescence test (Von dem Borne et al. 1980). Bone marrow variables studied included cellularity, number of megakaryocytes, the presence of morphological features of immaturity in the megakaryocytes, percentage of lymphocytes, plasma cells and eosinophils and iron stores. Particular attention was paid to the ITP treatments used before surgery. We analysed the use of prednisone alone
251
versus prednisone plus other treatments (other immunosuppressive agents, danazol, i.v. globulins), the number of different types of therapies used, the number of months on these therapies as well as the types of responses achieved by these treatments. The relapse rate after splenectomy was estimated by life-table analysis (Kaplan-Meiermethod) (KapIan & Meier. 1958). The remission duration was measured from the date of splenectomy to the date of relapse. Patients who were lost to follow-up are censored observations at the point in time when they were last seen in remission. The prognostic significance of the pre-splenectomy clinical variables was evaluated initially by univariate analysis (chisquare and t-test). When groups were ordered, a significant trend in proportions was determined by testing components of the overall chi-square statistic. The variables found to be significantly related to the achievement of long-lasting complete remissions were subsequently assessed in a multivariate proportional-hazards model, as described by Cox ( I 972). By means of a discriminant analysis we investigated the predictive capacity of the early (days 1-3) post-operative platelet counts. We used an analysis of variance to examine the relationship between maximum platelet count within the first 10 post-operative days and prognosis. The cases were grouped for comparison into three categories: less than 400 x 1OY//1, 400-600 x 1OY/1 and > 600 x 10Y/lplatelets. Calculations were computed by the BMDP program (Dixon & Brown, 1977). The response criteria to pre-splenectomy treatments and to splenectomy were the following: Complete remission was considered to be more than 150 x l O 9 / 1 platelets and no further treatment: complete remission with relapse, was more than 150 x 10Y//lplatelets, after day + 30 post-splenectomy, and subsequent relapse, and failure which includes lack of initial response, responses that require additional treatment or responses below the levels aforementioned. These working criteria were chosen from the guidelines proposed by Pizzuto (Pizzuto& Ambriz, 1984) but with minor modifications. To facilitate the statistical analysis of the data, we grouped the latter two categories of patients. In this manner, there were two groups of comparison: responders, made up of sustained complete remissions not requiring further treatment and non-responders, comprising all other types of response.
RESULTS
Patients Table I provides a summary of the main clinical features of the study population. Only 1 7 of 138 patients (12%) did not present with haemorrhagic manifestations. The clinical features related to the bleeding diathesis are also summarized in the table. If one accepts the persistence of disease for more than 6 months as the arbitrary dividing line between acute and chronic forms of ITP, then only 18 patients ( 13%) had the acute form of the disease. This is consistent with the age of the patients, since the chronic form is the one normally seen in the adult population. Our series included only seven patients
252
A. Julia et a1
Table I. Characteristics and bleeding manifestations in 1 3 8 splenectomized patients with chronic idiopathic thrombocytopenicpurpura n
Table 11. Platelet counts after splenectomy in patients with idiopathic thrombocytopenic purpura Platelets ( x
%
loy/])
-
Gender (F/M) Age (years)(mean fSD) Pre-Spx platelet count I 3 5 ( x I OY/I) Antiplatelet antibodies positive Interval diagnosis/splenectomy (months)(mean&SD) Bleeding manifestations
90/48 35f-18 132 26/38
96 68
20f31 121
88
Severity
Mild Moderate Severe
Petechiae Bruising Gingival Epistaxis Conjunctival Other sites of bleeding*
41 50 30 75 33 46 37 5 43
54 24 33 27 4 31
* 22 menorrhagia. 10 gastrointestinal. six urinary tract, five excessive surgical bleeding.
under 12 years of age. These paediatric cases were included because they had chronic thrombocytopenia (median diagnosis to splenectomy interval of 26 months). Bone marrow aspirates were available in 13 1 cases. Apart from the findings considered typical of the diagnosis (increased number of megakaryocytes. immaturity and little platelet production), we specifically examined the proportions of lymphocytes, plasma cells and eosinophils, that occasionally have been noted to be abnormal (Karpatkin, 1980). We found no differences whatsoever compared with the normal percentages (data not shown). Platelet antibodies were determined in a minority of patients. The percentage of positive titres (68%) and the types of immunoglobulins (16 IgG and seven IgM) can also be considered typical of patients with chronic ITP (Hegde et al, 1985). The mean diagnosis-to-splenectomy interval was 20 i3 1 months (range 1-242). Nine patients were splenectomized between 1 and 3 months from diagnosis and nine between 4 and 6 months. Forty-five patients in complete remission after splenectomy had been lost to follow-up at variable intervals since the operation (median 33 months, range 5-169 months). These cases have been treated as censored observations in the life-table analysis. The overall follow-up period was 48 f42 months with 2 5.4%of the patients controlled for periods exceeding 5 years. For the whole series, the median follow-up period was 36 months. Two patients (aged 5 1 and 64 years), nonresponders to splenectomy, died as a result of uncontrolled bleeding, 5 5 and 9 1 months respectively after the operation. All patients except five who were splenectomized at the outset, were previous therapeutic failures before undergoing splenectomy. 130 patients had thrombocytopenia that had been inadequately controlled with a tolerable dose of cortico-
Days post-operation
Mean
SD
1-3 7-10 20-40 90 180 360
168 438 2 50 225 225 226
145 316 21 1 146 146 140
steroids (prednisone) and three were refractory to other treatments (azathioprine). 20 patients who initially achieved complete remission on prednisone subsequently relapsed.
Response to splenectomy After splenectomy there was an immediate increase of the platelet count which was maximal around the tenth postoperative day and levelled-off thereafter (Table 11). As stated previously, 138 patients were treated with splenectomy. Of these, 91 achieved a complete remission without recurrence during the time of follow-up. The remaining 4 7 cases have been classified as treatment failures, including 23 with an early satisfactory response but subsequent relapse. The time of these recurrences ranged from the first to the seventh year after the operation. There was total lack of response in 1 6 cases (12%) and the remaining eight cases were other types of failures (suboptimal responses or partial remissions). In summary, the overall immediate complete remission rate was 83%: 114 of 138 operations, which includes 9 1 complete remissions and 23 complete remissions with subsequent relapse. This figure agrees with the 70-90% immediate response rates published in the literature (Doan et al, 1960: Picozzi et al, 1980; McMillan, 1981; Pizzuto & Ambriz, 1984; McVerry, 1985). The probability of staying in complete remission after splenectomy was estimated by the Kaplan-Meier method. The cumulative percentage of patients in complete remission is shown in Fig 1. Five years post-splenectomy, 57% of the patients are still in the complete remission achieved by the operation. Of the 23 patients that relapsed after being in complete remission, 11 (48%) did so within the first 6 months. Worthy ofnote is the fact that three patients relapsed more than 3 years after the operation. Identijication of clinical risk factors All the presplenectomy variables were examined by univariate analysis in relation to the achievement (91 patients) or not (47 patients) of a sustained complete remission. Two quantitative and three qualitative variables were found to be significantly related to the outcome in this one-dimensional analysis (P
Lack of useful clinical predictors of response to splenectomy in patients with chronic idiopathic thrombocytopenic purpura A. J U L I A , ’ c. ARAGUAS,’ J. ROSSELLO,’ J. B U E N O , ‘ P. D O M E N E C H , 2 M. O L O N A , ’ R. G U A R D I A , 3 J. P E T I T 4 AND A. FLORESS ‘Hospital ‘Val1d’Hebron’, Barcelona, 2Hospital ‘Princeps d’Espanya’, Barcelona, Hospital ‘Alvarez de Castro’, Gerona, * ‘Mutua de Tarrasa’, Barcelona, Hospital ‘Germans Trias’, Badalona, Spain
Received 20 December 7 989; accepted for publication 30 May 1990
Summary. We set out to identify clinical or analytical variables that might predict the response to splenectomy in patients with chronic idiopathic thrombocytopenic purpura (ITP). We retrospectively examined 138 mostly adult patients with chronic ITP, treated with splenectomy. Information was compiled from five Public Health Hospitals from a questionnaire and chart review. Sixty-one potentially prognostic variables were analysed by univariate and multivariate statistical methods. After multivariate analysis, age (relative risk=1.02; CI 1-1.03) and a severity of the bleeding
diathesis (relative risk= 1.6; CI 1.1 3-2.22) were independent prognostic factors for a sustained response to splenectomy. An independent analysis of the postsplenectomy counts showed that an early (days 1-3) thrombocyte count exceeding 156 x 10y/lcells increases the likelihood of a permanent unmaintained response. Our data indicate that the response to splenectomy in patients with chronic ITP cannot be adequately predicted on the basis of pre-splenectomy clinical or analytical variables.
Chronic idiopathic thrombocytopenic purpura (ITP) is a disorder in which platelets are destroyed by macrophages mostly in the spleen. In contrast to the acute disorder, chronic ITP rarely resolves spontaneously and requires the choice of one of several therapeutic options. Splenectomy removes the major site of platelet destruction (Gernsheimer et al, 1989) and with corticosteroids and immunosuppressive agents constitutes the group of the three most effective forms of treatment (Pizzuto & Ambriz, 1984). Splenectomy is usually indicated if a trial of steroids has failed or if the dose requirements are such that the patient is developing serious side effects. While a general consensus exists on the indications for splenectomy (Karpatkin, 1980: McVerry, 1985),it is unfortunate that the possible benefits of the operation seem to be totally unpredictable. Several groups have tried to detect useful predictors of the effect of the operation. While many workers agree upon the unpredictability of the response to splenectomy, others have reported on variables that might help in the prognosis of the disease after the operation. Those factors more frequently mentioned as predictive ones have been: age (the younger the better)
(Aster, 1972a: Di Fino et al, 1980; Akwari et al, 1987: Fenaux ct al, 1989; Fabris et al, 1989), previous response of the disease to prednisone (splenectomy more likely to be effective) (Thompson et aI, 1972: Jiji et nl, 1973: Brennan et al, 1975; Lacey & Penner, 1977; Mintz et al, 1981; Kernoff & Malan, 1983), short diagnosis-to-splenectomy interval (shorter intervals more likely to be followed by good surgical results), shortened platelet half-lives (shorter survivals predictive of better responses) (Siegel et al, 1989) and predominantly splenic sequestration (predictive of a good response) (Najean & Ardaillou, 1971; Gugliotta et al, 1981; Goudemand e t al, 1982; Fenaux et al, 1989; Gernsheimer et al, 1989). In contrast with these reports, the following references are from studies that failed to confirm the predictive value of the same five types of prognostic factors: age (Gugliotta et al, 1981: Goudemand et al, 1982), prednisone sensitivity (Doan et al, 1960; Gugliotta et al, 1981; Goudemand et al, 1982; Fenaux et al, 1989). time to splenectomy (Gugliotta et aI, 1981; Goudemand et al, 1982; Pizzutto & Ambriz, 1984; Fenaux et al, 1989), platelet survival (Gugliottaetal, 1981; Akwarietal, 1987;Fenauxetal, 1989)and site of sequestration (Aster, 1972b; Siegel et al, 1989). The aim of this study was to see whether a comprehensive analysis of the pre-splenectomy clinical and analytical features might help to identify those patients more likely to
Correspondence: Dr A. Julia, Servicio de Hematologia. Hospital General ‘Vall d’Hebron’,Po Vall d’Hebron, sin 08035. Barcelona. Spain.
2 50
Splniectomy in Patients with Chronic ITP respond. This report describes the treatment for 138 splenectomized patients and the statistical analysis of 61 potential prognostic variables. MATERIALS A N D METHODS This was a retrospective multicentre study with patients seen at five participating institutions. Details of each patient’s history and laboratory data were retrieved from hospital records and registered on a previously designed sheet. The diagnosis of ITP was established according to criteria generally accepted: thrombocytopenia with abundant or adequate megakaryocytes in bone marrow aspirate and absence of systemic disease, splenomegaly, coagulation disorder or any other likely cause of a low platelet count (Karpatkin, 1980: McMillan. 198 1). We evaluated a total of 6 1pre-splenectomy potentially prognostic clinical and analytical variables. The clinical variables included age, sex, year of diagnosis and splenectomy. time with haemorrhagic manifestations, interval between diagnosis and splenectomy and presence of coexisting diseases not aetiologically related to autoimmune thrombocytopenia. The characteristics of the bleeding diathesis were carefully recorded. In addition to noting the presence of petechiae, bruising, epistaxis, gingival or conjunctival bleeding and other sites of bleeding, for the purposes of this investigation we graded the two more common types of bleeding into four groups. Petechiae were grouped into none, less than 10, 10-100 or more than 100 lesions. Likewise, bruising was stratified into none, less than five haematomas, five to 10 lesions and more than 10. We also grouped the haemorrhagic manifestations into three categories: skin alone, mucosal alone or skin and mucosal. The overall intensity of the bleeding was considered to be absent, mild. moderate or severe, according to a generated score: mild bleeding was defined as the presence of skin manifestations alone (few petechiae, mild bruising); moderate bleeding as the association of more intense skin lesions with one mucosal bleeding site, and severe bleeding, as the combination of severe skin manifestations plus more than one site of mucosal haemorrhage. We analysed the following laboratory data: haematocrit, haemoglobin. platelet count, white blood cell count, absolute neutrophil count, lymphocyte, monocyte and eosinophil counts, at diagnosis, 3 months and 1 month prior to and immediately before splenectomy. A variable called ‘lowest’ (lowest platelet count of all pre-splenectomy determinations) was also independently evaluated. Post-operative peripheral blood counts were also recorded to assess the response at several time intervals, until the last known follow-up visit. Other pre-splenectomy variables included bone marrow specimen analysis, and presence and type of antiplatelet antibodies. Platelet antibodies were determined by an immunofluorescence test (Von dem Borne et al. 1980). Bone marrow variables studied included cellularity, number of megakaryocytes, the presence of morphological features of immaturity in the megakaryocytes, percentage of lymphocytes, plasma cells and eosinophils and iron stores. Particular attention was paid to the ITP treatments used before surgery. We analysed the use of prednisone alone
251
versus prednisone plus other treatments (other immunosuppressive agents, danazol, i.v. globulins), the number of different types of therapies used, the number of months on these therapies as well as the types of responses achieved by these treatments. The relapse rate after splenectomy was estimated by life-table analysis (Kaplan-Meiermethod) (KapIan & Meier. 1958). The remission duration was measured from the date of splenectomy to the date of relapse. Patients who were lost to follow-up are censored observations at the point in time when they were last seen in remission. The prognostic significance of the pre-splenectomy clinical variables was evaluated initially by univariate analysis (chisquare and t-test). When groups were ordered, a significant trend in proportions was determined by testing components of the overall chi-square statistic. The variables found to be significantly related to the achievement of long-lasting complete remissions were subsequently assessed in a multivariate proportional-hazards model, as described by Cox ( I 972). By means of a discriminant analysis we investigated the predictive capacity of the early (days 1-3) post-operative platelet counts. We used an analysis of variance to examine the relationship between maximum platelet count within the first 10 post-operative days and prognosis. The cases were grouped for comparison into three categories: less than 400 x 1OY//1, 400-600 x 1OY/1 and > 600 x 10Y/lplatelets. Calculations were computed by the BMDP program (Dixon & Brown, 1977). The response criteria to pre-splenectomy treatments and to splenectomy were the following: Complete remission was considered to be more than 150 x l O 9 / 1 platelets and no further treatment: complete remission with relapse, was more than 150 x 10Y//lplatelets, after day + 30 post-splenectomy, and subsequent relapse, and failure which includes lack of initial response, responses that require additional treatment or responses below the levels aforementioned. These working criteria were chosen from the guidelines proposed by Pizzuto (Pizzuto& Ambriz, 1984) but with minor modifications. To facilitate the statistical analysis of the data, we grouped the latter two categories of patients. In this manner, there were two groups of comparison: responders, made up of sustained complete remissions not requiring further treatment and non-responders, comprising all other types of response.
RESULTS
Patients Table I provides a summary of the main clinical features of the study population. Only 1 7 of 138 patients (12%) did not present with haemorrhagic manifestations. The clinical features related to the bleeding diathesis are also summarized in the table. If one accepts the persistence of disease for more than 6 months as the arbitrary dividing line between acute and chronic forms of ITP, then only 18 patients ( 13%) had the acute form of the disease. This is consistent with the age of the patients, since the chronic form is the one normally seen in the adult population. Our series included only seven patients
252
A. Julia et a1
Table I. Characteristics and bleeding manifestations in 1 3 8 splenectomized patients with chronic idiopathic thrombocytopenicpurpura n
Table 11. Platelet counts after splenectomy in patients with idiopathic thrombocytopenic purpura Platelets ( x
%
loy/])
-
Gender (F/M) Age (years)(mean fSD) Pre-Spx platelet count I 3 5 ( x I OY/I) Antiplatelet antibodies positive Interval diagnosis/splenectomy (months)(mean&SD) Bleeding manifestations
90/48 35f-18 132 26/38
96 68
20f31 121
88
Severity
Mild Moderate Severe
Petechiae Bruising Gingival Epistaxis Conjunctival Other sites of bleeding*
41 50 30 75 33 46 37 5 43
54 24 33 27 4 31
* 22 menorrhagia. 10 gastrointestinal. six urinary tract, five excessive surgical bleeding.
under 12 years of age. These paediatric cases were included because they had chronic thrombocytopenia (median diagnosis to splenectomy interval of 26 months). Bone marrow aspirates were available in 13 1 cases. Apart from the findings considered typical of the diagnosis (increased number of megakaryocytes. immaturity and little platelet production), we specifically examined the proportions of lymphocytes, plasma cells and eosinophils, that occasionally have been noted to be abnormal (Karpatkin, 1980). We found no differences whatsoever compared with the normal percentages (data not shown). Platelet antibodies were determined in a minority of patients. The percentage of positive titres (68%) and the types of immunoglobulins (16 IgG and seven IgM) can also be considered typical of patients with chronic ITP (Hegde et al, 1985). The mean diagnosis-to-splenectomy interval was 20 i3 1 months (range 1-242). Nine patients were splenectomized between 1 and 3 months from diagnosis and nine between 4 and 6 months. Forty-five patients in complete remission after splenectomy had been lost to follow-up at variable intervals since the operation (median 33 months, range 5-169 months). These cases have been treated as censored observations in the life-table analysis. The overall follow-up period was 48 f42 months with 2 5.4%of the patients controlled for periods exceeding 5 years. For the whole series, the median follow-up period was 36 months. Two patients (aged 5 1 and 64 years), nonresponders to splenectomy, died as a result of uncontrolled bleeding, 5 5 and 9 1 months respectively after the operation. All patients except five who were splenectomized at the outset, were previous therapeutic failures before undergoing splenectomy. 130 patients had thrombocytopenia that had been inadequately controlled with a tolerable dose of cortico-
Days post-operation
Mean
SD
1-3 7-10 20-40 90 180 360
168 438 2 50 225 225 226
145 316 21 1 146 146 140
steroids (prednisone) and three were refractory to other treatments (azathioprine). 20 patients who initially achieved complete remission on prednisone subsequently relapsed.
Response to splenectomy After splenectomy there was an immediate increase of the platelet count which was maximal around the tenth postoperative day and levelled-off thereafter (Table 11). As stated previously, 138 patients were treated with splenectomy. Of these, 91 achieved a complete remission without recurrence during the time of follow-up. The remaining 4 7 cases have been classified as treatment failures, including 23 with an early satisfactory response but subsequent relapse. The time of these recurrences ranged from the first to the seventh year after the operation. There was total lack of response in 1 6 cases (12%) and the remaining eight cases were other types of failures (suboptimal responses or partial remissions). In summary, the overall immediate complete remission rate was 83%: 114 of 138 operations, which includes 9 1 complete remissions and 23 complete remissions with subsequent relapse. This figure agrees with the 70-90% immediate response rates published in the literature (Doan et al, 1960: Picozzi et al, 1980; McMillan, 1981; Pizzuto & Ambriz, 1984; McVerry, 1985). The probability of staying in complete remission after splenectomy was estimated by the Kaplan-Meier method. The cumulative percentage of patients in complete remission is shown in Fig 1. Five years post-splenectomy, 57% of the patients are still in the complete remission achieved by the operation. Of the 23 patients that relapsed after being in complete remission, 11 (48%) did so within the first 6 months. Worthy ofnote is the fact that three patients relapsed more than 3 years after the operation. Identijication of clinical risk factors All the presplenectomy variables were examined by univariate analysis in relation to the achievement (91 patients) or not (47 patients) of a sustained complete remission. Two quantitative and three qualitative variables were found to be significantly related to the outcome in this one-dimensional analysis (P
