Long-term recurrence and mortality after surgery and adjuvant radiotherapy for nonfunctional pituitary adenomas

736 J. Neurosurg. / Volume 108 / April 2008

NDOCRINE-INACTIVE pituitary adenomas (or NFPAs)are the most common presenting pituitary tumor.The overall population prevalence of NFPAs de-

rived from pathological and imaging studies has been esti-mated to be up to 16.7%.13,19,27 In contrast to other pituitary

adenomas characterized by hormonal excess, NFPAs pre-sent with symptoms that typically arise from growth-relat-ed mass effect on surrounding structures. Visual field dis-turbances and hormonal deficiency reflect damage fromcompression of the optic chiasm and anterior pituitary lobe,respectively. Hyperprolactinemia can occur from compres-sion of the pituitary gland or stalk and subsequent loss oftonic dopaminergic inhibition of prolactin secretion fromthe normal gland (known as stalk effect). Surgical removalis the first definitive step in the treatment of symptomaticNFPAs.

Despite their high prevalence, the slow growth of NFPAs

J Neurosurg 108:736–745, 2008

Long-term recurrence and mortality after surgery andadjuvant radiotherapy for nonfunctional pituitary adenomas

EDWARD F. CHANG, M.D.,1 GABRIEL ZADA, M.D.,2 SANG KIM, M.D.,3

KATHLEEN R. LAMBORN, PH.D.,1,4 ALFREDO QUINONES-HINOJOSA, M.D.,5

J. BLAKE TYRRELL, M.D.,6 CHARLES B. WILSON, M.D.,1 AND SANDEEP KUNWAR, M.D.1

Departments of 1Neurological Surgery, 4Biostatistics, and 6Medicine, Division of Endocrinology,University of California, San Francisco; 2Department of Neurological Surgery, Universityof Southern California, Los Angeles, California; 3Department of Medicine, Harvard MedicalSchool, Boston, Massachusetts; and 5Department of Neurological Surgery, Johns HopkinsUniversity School of Medicine, Baltimore, Maryland

Object. Long-term outcomes following surgery for nonfunctional pituitary adenomas (NFPAs) are unclear. The roleof adjuvant radiation therapy is therefore controversial because it is associated with higher tumor control but also car-ries known long-term morbidity. The authors’ aim was to determine predictors of recurrence and overall survival andto define patient subgroups that may benefit from radiotherapy.

Methods. The authors performed a retrospective cohort analysis of 663 patients who underwent surgery between1975 and 1995 for treatment of primary NFPAs. The main outcome measures were disease progression after surgery,defined by clinical and/or imaging criteria, and all-cause mortality.

Results. Over a median clinical follow-up of 8.4 years, there were 64 (9.7%) recurrences after treatment, with a medi-an time to recurrence of 5.6 years. The 5-, 10-, and 15-year recurrence-free probabilities were 0.93, 0.87, and 0.81, re-spectively. Multivariate Cox proportional hazard regression analysis identified the following predictors as associatedwith increased recurrence: cavernous sinus invasion (hazard ratio [HR] 3.6, 95% confidence interval [CI] 1.5–6.4; p ,0.001) and subtotal resection (STR) without radiotherapy (HR 3.6, 95% CI 1.4–14; p = 0.01). Using time-to-event esti-mates to adjust for differences in follow-up between groups, radiotherapy was found to reduce tumor recurrence in onlythose patients who received an STR (p , 0.001, log-rank test) but not gross-total resection (GTR; p = 0.63, log-ranktest). Median follow-up for overall survival was 14.0 years. The 5-, 10-, 15- and 20-year overall survival estimates were0.91, 0.81, 0.69, and 0.55, respectively. Within the study cohort and in age- and sex-adjusted comparison with the gen-eral US population, increased relative mortality was observed in patients who underwent radiotherapy or STR.

Conclusions. Cavernous sinus invasion is an important prognostic variable for long-term control of NFPAs. Ra-diotherapy results in long-term tumor control for patients who undergo STR but does not affect recurrence rates andmay increase the risk of death after GTR. Given the risks associated with radiotherapy, there is no role for its rou-tine application in patients who have undergone GTR of their NFPA. In all patients, long-term monitoring is required.(DOI: 10.3171/JNS/2008/108/4/0736)

KEY WORDS • endocrine-inactive pituitary adenoma • nonfunctional pituitary adenoma •radiation therapy • recurrence • survival • transsphenoidal surgery

EAbbreviations used in this paper: ACTH = adrenocorticotropic

hormone; CI = confidence interval; CT = computed tomography;GH = growth hormone; GTR = gross-total resection; HR = hazardratio; MR = magnetic resonance; NDI = National Death Index;NFPA = nonfunctional pituitary adenoma; OR = odds ratio; SSDI =Social Security Death Index; STR = subtotal resection.

has limited the number of studies that adequately describethe long-term recurrence and mortality rates after surgery.Furthermore, studies have demonstrated highly variableoutcomes. Authors of some recent studies have advocat-ed the universal use of radiotherapy after surgery becauseradiotherapy results in higher control rates than surgeryalone.15,17 Recurrence rates have been estimated at 50–60%at 10 years after surgery without irradiation,36 but only 7%after combined surgery and radiotherapy.17 Observation-al studies, however, have discovered associations betweenradiotherapy and hypopituitarism as high as 50%,34 andin some reported cases, neurocognitive impairment,29 opticneuropathy,5,14 and secondary intracerebral oncogenesis.4,26

Other studies, however, have demonstrated high controlrates after surgery without radiotherapy.10,18,22 As a result,the specific role of radiotherapy after surgery remains high-ly controversial.16

We therefore conducted a large retrospective cohortstudy of patients who underwent surgery for NFPAs over a20-year period with long-term follow-up. The purpose ofthis study was to describe the predictors of initial extent ofresection and to determine the long-term recurrence andmortality rates after surgery, with and without radiotherapy,for NFPAs.

Clinical Materials and Methods

Study Design and Population

We conducted a retrospective review of the hospital andoffice records of 663 consecutive patients who underwentresection of primary NFPAs at the University of California,San Francisco. Operations were performed exclusively by1 surgeon (C.B.W.). The records of all patients who under-went surgery between January 1975 and December 1995were reviewed. A cutoff date of December 31, 1995, wasused to ensure adequate follow-up.

Patients in this study were limited to those who under-went surgery at the University of California, San Francisco,as the primary definitive treatment for their NFPA. “Non-functional” refers to the clinical behavior of these adenomasand not necessarily their capacity for producing hormonesor their immunohistochemical staining. The diagnosis ofNFPA was based on the patient’s clinical history and pre-sentation, endocrinological status, imaging findings, andhistopathological findings. Elevated prolactin was consid-ered stalk effect when the level was less than would be ex-pected for size of tumor.

Patients were excluded from the study if they had under-gone any previous pituitary surgery or had a concurrentpituitary condition (such as a hormone-secreting tumor,craniopharyngioma, or Rathke pouch cyst).

Clinical and Neuroimaging Evaluation

Immediately before surgery all patients were interviewedto obtain the details of their illness, and a complete physi-cal and neurological examination was performed. An en-docrinological evaluation was usually performed preopera-tively, including sampling for prolactin, thyroid-stimulatinghormone, thyroxine, ACTH, testosterone, luteinizing hor-mone, follicle-stimulating hormone, GH, and insulin-likegrowth factor–1 when available.

The method of preoperative imaging diagnosis of pitu-itary tumors evolved over the 20-year study period. Initialimaging evaluation consisted of CT scanning, which wassuperceded by MR imaging in sagittal and coronal planes,with and without Gd enhancement. Magnetic resonance im-aging remains the current standard for evaluating the stateof the pituitary and sella turcica.

Surgical Procedure

The majority of surgeries were performed via the sub-labial transsphenoidal approach. The general indicationsfor transsphenoidal removal of symptomatic NFPAs werethe following: 1) extrasellar extension, including extensioninto the sphenoid sinus; 2) pituitary apoplexy; 3) intract-able headaches; 4) visual impairment; or 5) pituitary insuf-ficiency. With the aid of intraoperative fluoroscopy, a sub-labial transsphenoidal approach to the sella was used. Theexact strategy for each resection was determined by the sur-geon at the time of operation and was based on tumor size,consistency, shape, and location. In each case, the goal wasthe selective removal of the tumor tissue while sparing nor-mal gland.

In certain situations, a transfrontal craniotomy was be-lieved to be safer and more effective than the transsphe-noidal approach. The indications for craniotomy were thefollowing: 1) minimal enlargement of the sella relative toa large suprasellar mass; 2) extrasellar expansion into themiddle fossa, with the volume of the extrasellar mass beinggreater than that of the intrasellar mass; 3) presence of anunrelated intracranial pathological entity that could com-plicate the transsphenoidal operation (for example, a para-sellar aneurysm); and 4) unusually fibrous tumor that couldbe removed transsphenoidally, in which case a craniotomywas undertaken as a second procedure to relieve compres-sion of the compromised optic chiasm.

Extent of resection (gross total or subtotal) was evalu-ated using postoperative images and also by the surgeon’sintraoperative impression. Of all patients, 72.8% under-went postoperative imaging in the 6 months following sur-gery. Another 16% underwent imaging between 6 monthsand 1 year following surgery. In almost all cases, we on-ly had access to imaging reports and clinic documentationof the imaging results and not the images themselves. Thesurgeon’s intraoperative impression was used in cases inwhich GTR was noted in the postoperative imaging reportbut the surgeon’s intraoperative impression was STR. Thisoccurred in 78 cases, 69 of which were based on CT scan-ning findings and 9 of which were based on MR imagingfindings.

Although patients were not randomized for postoper-ative adjuvant radiotherapy, changes in the managementstrategy during the course of the study period facilitatedcomparison between patient cohorts that did and did notreceive radiotherapy. Previous to 1987, STRs occurred at ahigher rate, and radiotherapy was usually applied routine-ly regardless of extent of resection given its known highcontrol rates. After 1987, because of our concerns of pitu-itary failure after radiotherapy, GTR with the preservationof normal gland became the surgical goal. The number ofGTRs after 1987 increased dramatically, and radiotherapywas applied more selectively. Some patients did not receiveradiotherapy despite STR, mainly because of preference to

J. Neurosurg. / Volume 108 / April 2008

Long-term outcomes for nonfunctional pituitary adenomas

737

avoid side effects, given young childbearing age or relative-ly intact anterior lobe function. As a result of these practicemodifications, the natural history for various subgroups(extent of resection and radiotherapy) could be assessed.

Radiation Therapy

Conventional radiotherapy permits a nearly uniformdose of radiation to be delivered to a carefully defined tu-mor volume, with a rapid reduction in the radiation outsidethe tumor volume resulting in maximum sparing of ad-jacent normal tissues. The preferred technique involves bi-lateral moving arc fields with reversing 30° wedge filters.Typically, daily treatments with 1.8 cGy are continued for5 days/week to a total dose of 45–54 cGy.

Classification of Tumors

Tumors were classified based on neuroimaging and in-traoperative findings of the degree of sellar destruction(grade) and extrasellar extension (stage), according to thegrading system developed by Hardy20 and modified by Wil-son.37 Other characteristics including tumor size, supra-sellar extension, sphenoid sinus invasion, cavernous sinusinvasion, presence of necrosis, cysts, or hemorrhage wererecorded.

Initially, surgical specimens were characterized by a tra-ditional classification scheme based on acidic and basicstaining affinities. After 1983, surgical pathology speci-mens were classified using contemporary criteria by rou-tine immunohistochemical processing by using antibodiesto GH, ACTH, and prolactin. The NFPAs were classifiedinto the following tumor subtypes: null cell, silent corti-cotroph, silent somatotroph, and silent lactotroph adeno-mas. Null cell adenomas were defined by staining nega-tively for anterior pituitary hormone production, includingprolactin, ACTH, and GH. Silent corticotropic adenomasstained positively for ACTH but were not associated withclinical or serum-based evidence of hypercortisolism. Asmall number of silent somatotroph and lactotroph adeno-mas were included, given their lack of clinical manifesta-tion of hormonal excess despite the secretion of a clinical-ly measurable product.

Follow-Up Period

Interim follow-up information was obtained through theretrospective review of hospital and departmental recordsand follow-up contact using protocols and questionnairesapproved by the University of California, San Francisco,research institutional review board. We attempted to con-tact all patients living in the US to obtain current clinicalinformation by mail-in questionnaire and telephone in-terview. Information was obtained regarding the clinicalstatus of each patient contacted, current medical therapy,and any intervening therapies, such as another operation orradiation treatment. All patients were asked if they hadtaken hormone replacement medications and if they expe-rienced symptoms of anterior pituitary dysfunction at anypoint following surgery. Particular emphasis was placedon obtaining information regarding tumor recurrence/pro-gression and regarding results of patients’ last MR imagingstudy.

Recurrence was usually diagnosed by follow-up neu-

roimaging and/or clinical symptoms. Recurrence was de-fined in this study as the progression of tumor growth thatresulted in another treatment, including repeated surgicalremoval or radiotherapy. This clinical criterion for recur-rence was used as opposed to neuroimaging documentationof adenoma regrowth because we believed it to be the mostobjective, sensitive, and clinically meaningful measure ofprogression. We only had access to postoperative imagingreports and not the images themselves. Furthermore, webelieved that patients in the follow-up period were far morereliable in recalling another procedure and dates than theresults and dates of their repeated imaging studies. Thenumber of patients who underwent CT scanning for fol-low-up imaging was 181, although 112 eventually under-went MR imaging once it was readily available in the early1980s. The routine follow-up for NFPAs involved serialMR images/CT scans at 3, 6, and 12 months, then yearlyuntil 5 years, then every 5 years thereafter.

Death events were obtained from medical records andsupplemented by the SSDI and NDI. The NDI and SSDIhave been used in other published studies to evaluate long-term mortality, with sensitivity of 93–97% and specificityof 100% (when social security numbers are known).11,32 Allknown and confirmed deaths from our institution’s hospitalrecords for patients in our database were also found in bothSSDI and NDI databases. The cause of death was unavail-able for the majority of patients. Follow-up data pertainingto death events were updated to January 1, 2005, which wasthe last date archived by the NDI and SSDI at the time ofdata request.

Statistical Analysis

Logistic regression was used to analyze predictors of ex-tent of resection. This was done for each variable individu-ally (univariate) and also using a backward, stepwise mul-tivariate regression procedure. Kaplan–Meier analysis wasused to estimate tumor recurrence and overall survival. Thelog-rank test was used to evaluate differences in outcomebetween patients who underwent a GTR versus an STRand those who received radiation treatment compared withthose who did not. Expected mortality rates by age and sex,divided into 5-year age groups, were obtained from the USBureau of Census and compared with those observed inthis study. The expected mortality rates for patients in thisstudy were determined based on person–years of follow-upfor each sex and age group. The difference between ob-served and expected rates was tested using the exact prob-abilities of the Poisson distribution. Results were consid-ered statistically significant if the two-tailed probabilityvalue was , 0.05.

Results

Patient Population and Clinical Presentation

Table 1 summarizes the baseline characteristics of thestudy population and provides details on surgical and ra-diotherapy subgroups. The median patient age was 53years (range 16–88 years), and 6 patients were , 18 yearsof age. There were 394 male (59%) and 272 female (41%)patients. Most of the presenting characteristics of patientswith NFPAs were consistent with mass effect causing com-

E. F. Chang et al.


pression of the optic pathways or gland. Visual loss was the most common presenting symptom (49%), followed by headache (32%). In male patients, complaints of de-creased libido were common (37%). For female patients,abnormal menses and/or galactorrhea occurred in 34%.Other common complaints were diplopia (3.9%) and “hy-pothyroid” constitutional symptoms (16%, including $ 2of the following symptoms: lethargy, constipation, weak-ness, weight gain, alopecia, or cold intolerance). Other lessfrequent symptoms included seizures, infertility, syncope,and cognitive deficits; 7.4% of NFPAs were discovered asincidental findings, usually through MR imaging in the sec-ond decade of the study.

Study population characteristics were compared be-tween the surgery alone (49%) versus surgery plus ra-diotherapy (51%) groups. The proportion of patients withpreoperative visual deficits was significantly higher in theradiotherapy group. The patients in the radiotherapy groupalso had longer follow-up periods, given that more patientsreceived radiotherapy in the early half of the study period.Age at surgery, sex, and symptoms other than visual defectswere not associated with the use of radiotherapy.

Adenoma Characteristics

Tumor features and operative details are summarized inTable 2. The median adenoma size at maximum diameterwas 25 mm (range 6–45 mm). As expected, the vast major-

ity of these tumors were macroadenomas (at least Grade II,as defined by a diameter . 10 mm), accounting for 96% ofall tumors in the study. The exact degree of sellar floordestruction was inconsistently recorded; therefore, GradesIII and IV (localized and diffuse, respectively) were com-bined. Sellar floor destruction with infrasellar extensionwas present in 26% of patients. Tumor stage for pituitaryadenomas refers to the degree of extrasellar extension. Su-prasellar extension was found in 71% of patients (StagesA–C). Cavernous sinus invasion was found in 58 patients(9%). Other characteristics noted at the time of surgeryincluded cystic features (9.4%), necrosis (3.5%), and hem-orrhage (5.7%).

Surgical specimens underwent immunostaining forprolactin, GH, and ACTH in 313 patients. The major-ity of NFPAs did not stain for any hormone marker andwere therefore considered null cell adenomas (88%). TheACTH-positive stained specimens, suggesting silent corti-cotroph adenomas, were found in 6.1% of patients. Otheradenomas also stained positively, albeit mostly weak forprolactin (4.8%) and GH (1.6%). Again, these hormonepositive–staining adenomas were classified as NFPAs giventhe lack of clinical symptoms or syndrome in the patientsharboring them.

Adenoma size was larger in the radiotherapy group thanin the surgery alone group (median size in the surgerygroup 23 mm versus combined surgery and radiotherapygroup, 30 mm, p , 0.001). This was further reflected in agreater proportion of patients with Grade II adenomas andalso a greater need for transfrontal surgical approach in thegroup that received radiotherapy (Table 1).

Extent of Resection

Of the 663 patients who underwent resection of primaryNFPAs during the study period, 307 patients (46%) re-ceived a GTR. Subtotal resection was performed in 356 pa-tients (54%).

A considerable shift in the operative results occurredaround 1987. Prior to 1987, only 33% of surgeries weredeemed GTRs. In contrast, after 1987, 57% of surgerieswere GTRs. This difference was highly significant (p ,0.001) and reflects the strong influence that operative goalsplayed in the excision of NFPAs. Again, in the early peri-od, the goal of surgery was general tumor debulking, whichwas followed by routine radiotherapy. Conversely, in thelater period, the surgical priority was to remove as muchtumor as possible, regardless of the need for radiotherapy.Therefore, to remove the influence of the surgeon’s ap-proach to the resection, only patients who underwent sur-gery between 1987 and 1995 were analyzed for predictorsof extent of resection (356 patients).

In univariate logistic regression analysis of potential pre-dictors of STR (Table 3), the factors associated with STRincluded hypopituitarism (p = 0.03), visual defects (p ,0.001), sphenoid sinus/infrasellar extension (p = 0.01), andcavernous sinus invasion (p , 0.001). Greater overall sizewas also associated with higher odds of STR, especially forNFPAs . 20 mm (p , 0.001). Patient age, sex, and histo-logic subtype were not significantly associated with extentof resection.

When these variables were entered into multivariate an-alysis, only cavernous sinus invasion (p , 0.001) and size



739

TABLE 1Baseline clinical characteristics in 663

patients who underwent resection of NFPAs

No. of Patients (%)

All Surgery Surgery & pPatients Alone Radiotherapy Value

total 663 323 340age (yrs)

median 53 53 53 0.64range 16–88 22–81 16–88

sex 0.38male 394 (59) 187 (58) 207 (61)female 272 (41) 136 (42) 133 (39)

symptoms at presentationvisual deficits 327 (49) 118 (37) 209 (61) ,0.001headache 212 (32) 102 (32) 110 (32) 0.45diplopia 26 (3.9) 12 (3.7) 14 (4.1) 0.78decreased libido (men) 146 (37) 71 (38) 75 (36) 0.95amenorrhea/galactorrhea 92 (34) 55 (40) 37 (28) 0.06(women)

constitutional hypo- 104 (16) 46 (14) 58 (17) 0.35thyroidism

apoplexy 24 (3.6) 10 (3.1) 14 (4.1) 0.47incidental finding 49 (7.4) 29 (9.0) 20 (5.9) 0.05

timing of surgery1975–1987 307 (46) 84 (26) 223 (66) ,0.0011988–1995 356 (54) 239 (74) 117 (34)

op approachtranssphenoidal 638 (96) 318 (98) 320 (94) 0.003transfrontal 25 (4) 5 (1.5) 20 (6)

median radiotherapy 45 dose in Gy (range) (39.60–55)

median follow-up (yrs)recurrence 8.4 6.8 9.8 ,0.001overall survival 14.0 12.8 17.1 0.12

(p = 0.003) were found to be significantly associated withSTR (Table 3). The NFPAs with cavernous sinus invasionhad almost 7 times the odds (OR 6.7, 95% CI 2.38–18.8) of receiving an STR. Each millimeter increase in adenomasize was also associated with an 11% increase in the oddsof STR (OR 1.11, 95% CI 1.06–1.15).

Tumor Progression and Recurrence

The median follow-up period for evaluation of tumorrecurrence was 8.4 years (663 patients, range 0–29.0 years;the period during which patients received serial postop-erative imaging and clinical evaluation). Of 663 patients,547 had follow-up for . 6 months, 456 for . 2 years, 350for . 5 years, and 208 patients for . 10 years. Tumor re-curred in 64 patients (9.7%), with an overall median timeto recurrence of 5.56 years. Patients who underwent GTR(307 patients) had 19 recurrences, of which 6 occurred inpatients who underwent radiotherapy and 13 in those whodid not. The median time to recurrence among these pa-tients was 5.3 and 6.9 years, respectively. Of the 356 pa-tients who underwent STRs, 45 had recurrences, of whom24 underwent radiotherapy and 21 did not prior to the re-currence. The median time to recurrence among these pa-tients was 8.7 and 3.8 years, respectively.

Kaplan–Meier estimates were calculated to adjust fordifferential durations of follow-up between groups. Withregard to extent of resection, recurrence rates were higherfor patients who received STR compared with GTR (p =0.02, log-rank test). When stratified separately for adjuvantradiotherapy, patients who underwent STR alone had sig-nificantly lower tumor-free survival rates than patients whounderwent STR and adjuvant radiotherapy (p , 0.005, log-rank test). In patients who underwent GTR, no differencewas observed between patients who had adjuvant radio-therapy and those who underwent surgery alone (p = 0.63,

E. F. Chang et al.


TABLE 2Tumor characteristics in 663 patients who underwent resection of NFPAs

No. of Patients (%)

All Surgery Alone Surgery & Radio-(663 patients) (323 patients) therapy (340 patients) p Value

size (mm)median 25 23 30 ,0.001*range 6–45 6–40 8–45

Hardy–Wilson ClassificationGrade: relationship of adenoma to sella & sphenoid sinuses 0.04†

I: sella normal or focally expanded; tumor ,10 mm 26 (3.9) 19 (5.9) 7 (2.1)II: sella enlarged; tumor .10 mm 467 (70) 231 (72) 236 (69)III/IV: localized or diffuse perforation of sellar floor 170 (26) 73 (23) 97 (29)

stage: extrasellar extensionsuprasellar

0 (no extension) 42 (6.3) 22 (6.8) 20 (5.9) 0.66‡A–C (suprasellar) 474 (71) 231 (72) 243 (71) 0.91

parasellar D (intracranial) 5 (0.8) 2 (0.6) 3 (0.9) 0.85E (cavernous sinus) 58 (9) 18 (5.6) 40 (12) 0.004

tumor featurecystic 62 (9.4) 32 (10) 30 (8.8) 0.64‡necrotic 23 (3.5) 22 (6.8) 11 (3.2) 0.59hemorrhagic 38 (5.7) 21 (6.5) 17 (5.0) 0.21

immunostaining§null cell adenoma 274 (88) 145 (83) 129 (93) 0.15‡ACTH-positive 19 (6.1) 15 (8.6) 4 (2.9)prolactin-positive 15 (4.8) 11 (6.3) 4 (2.9)GH-positive 5 (1.6) 3 (1.7) 2 (1.4)

* Unpaired t-test.† Wilcoxon signed-rank test.‡ Chi-square test.§ Immunostaining findings were available in 313 patients (174 [56%] in the surgery group and 139 [44%] in the surgery and radio-

therapy groups).

TABLE 3Predictors of subtotal resection: uni- and multivariate

logistic regression analyses

Logistic Regression Analyses OR (95% CI) p Value

univariateage 0.99 (0.99–1.01) 0.32female 1.3 (0.93–1.73) 0.31preop prolactin 0.99 (0.99–1.01) 0.40

any hypopituitarism 1.37 (1.01–1.87) 0.03visual defect 4.09 (2.60–6.41) ,0.001suprasellar extension 1.48 (0.87–2.52) 0.14sphenoid sinus invasion 2.27 (1.21–4.27) 0.01cavernous sinus invasion 5.9 (2.64–13.56) ,0.001cystic 0.93 (0.48–1.78) 0.83necrosis 0.56 (0.17–1.83) 0.34hemorrhage 0.51 (0.21–1.23) 0.14size (mm) 1.1 (1.06–1.15) ,0.001

multivariatecavernous sinus invasion 6.70 (2.38–18.8) ,0.001size (mm) 1.11 (1.06–1.15) 0.003

log-rank test). No difference was detected for patients whohad a GTR and those who had an STR and subsequentradiotherapy (p = 0.43, log-rank test; Fig. 1).

These observations were confirmed using a Cox pro-portional hazard regression analysis. Univariate analysisshowed that cavernous sinus invasion, STR, and no ra-diotherapy were associated with increased recurrence (Ta-ble 4). Of note, silent corticotroph NFPAs did not showdemonstrable association with increased recurrence (p =0.978). The significant univariate factors were entered in-to the multivariate analysis. A clear difference in the re-currence rate of patients with STRs without radiotherapyshown in the Kaplan–Meier survival curves led us to in-troduce an interaction term to test for confounding effectsbetween extent of resection and the administration of radio-therapy. Cavernous sinus invasion (HR 3.61, 95% CI 1.49–6.37, p , 0.001) and the specific combination of STR andno radiotherapy (HR 3.57, 95% CI 1.39–13.9, p = 0.01)jointly predicted NFPA recurrence (Table 4). Meanwhile,STR (HR 1.05, 95% CI 0.42–2.59, p = 0.91) and no radio-therapy (HR 1.19, 95% CI 0.45–3.19, p = 0.72) did not ap-pear to be independent significant predictors. The recur-rence-free survival probabilities are shown in Table 5.

Given the strong influence that cavernous sinus invasionplayed in predicting extent of resection and the administra-tion of radiotherapy, adjustment was made for this furtherpotential confounding factor. A separate analysis includingonly those patients without cavernous sinus invasiondemonstrated that STR and no radiotherapy were indepen-dently predictive of NFPA recurrence (STR: HR 3.71, 95%CI 1.78–7.69, p , 0.001; no radiotherapy: HR 3.96, 95%CI 1.95–8.06). Together, both log-rank and proportionalhazard analyses suggest that patients with cavernous sinusinvasion and those who receive an STR without radiother-apy are at increased risk of tumor recurrence.

Mortality Rate

Overall, 380 patients (57%) died during the follow-upperiod (up to January 1, 2005). The median follow-upusing both hospital charts and national death indices was14.0 years. The median overall survival for all patients was81% at 10 years and 55% at 20 years. Initial assessment ofoverall survival difference between treatment-related vari-ables (such as extent of resection and radiotherapy) trend-ed toward, but did not reach, statistical significance (p =0.06, log-rank test; Fig. 2A).

However, on plotting overall survival curves for eachgroup, it became apparent that significant differences inmortality rates had occurred later in the follow-up period.Subanalysis for late deaths occurring . 10 years after sur-gery revealed decreased mortality rates in patients who un-derwent GTRs without irradiation compared with othergroups (STR with and without radiotherapy and GTR withradiotherapy) (p = 0.048, log-rank test; Fig. 2B). See Table6 for overall survival estimates.

The univariate Cox proportional hazard demonstratedthat older age at surgery, larger tumor size, male sex, STR,and radiotherapy were associated with decreased survival(Table 7). These variables were entered into a multivari-ate model, also including an interaction term for those pa-tients who underwent a GTR and no radiotherapy (based on

the survival plots). Age at surgery, as expected, predictedshorter survival (HR 1.09, 95% CI 1.08–1.84, p , 0.0001).The combination of a GTR and no radiotherapy converse-ly showed longer survival and decreased mortality rates(HR 0.72, 95% CI 0.54–0.94, p = 0.02; Table 7). Extent ofresection or radiotherapy alone did not appear to have inde-pendent effects on overall survival (HR 0.94, 95% CI0.62–1.4, p = 0.76).

Given these results, the probability of death for patientsin each group with the number of deaths expected basedon age- and sex-matched population survival statisticswas also evaluated. Overall, the total study population hada higher mortality rate than the general population (p ,0.001). Subgroup analysis demonstrated that patients who



741

TABLE 4Predictors of NFPA recurrence:

uni- and multivariate Cox regression

Logistic Regresssion Analysis HR (95% CI) p Value

univariateage 0.97 (0.97–1.0) 0.79size .25 mm 2.32 (0.79–6.8) 0.127extension

suprasellar 1.15 (0.37–3.6) 0.872infrasellar 0.73 (0.22–2.4) 0.605cavernous sinus invasion 3.44 (1.3–9.3) 0.014

pathology: null cell adenoma 1.0 (0.0–105) 0.978pathology: silent corticotroph 1.0 (0.0–105) 0.978STR 3.1 (1.1–10) 0.049no radiotherapy 6.21 (1.9–20) 0.002

multivariate cavernous sinus invasion 3.61 (1.49–6.37) ,0.001STR 1.05 (0.42–2.59) 0.91no radiotherapy 1.19 (0.45–3.19) 0.72STR & no radiotherapy 3.57 (1.39–13.9) 0.01

FIG. 1. Kaplan–Meier plot of recurrence-free survival after resec-tion in 663 patients with NFPAs. Patients who underwent an STRwithout radiotherapy had far shorter time to recurrence than all oth-er groups (p , 0.005, log-rank test). No statistical difference wasobserved between patients who underwent GTR with and withoutradiotherapy (p = 0.43, log-rank test).

underwent STR or radiotherapy had higher observed mor-tality (p = 0.001 and p , 0.001, respectively), whereas inpatients who underwent GTR and/or did not receive radio-therapy the result was less pronounced (p = 0.06 and p =

0.11, respectively). Taken together with direct group com-parisons, these findings suggest that patients who under-went GTR without radiotherapy had nearly normal overallsurvival, compared with shorter overall survival for thosewith an STR alone and those who received radiotherapy.

Four patients died during the perioperative period (allafter transsphenoidal surgery) for an overall procedure-re-lated mortality rate of 0.2%. Two patients died of meningi-tis despite the administration of intravenous antibiotics, and1 each of pulmonary embolism and intracerebral hemor-rhage.

Discussion

This study is a retrospective analysis of a large cohort ofpatients with long-term follow-up after surgery for NFPAs.Our study confirms that transsphenoidal resection is a safeprocedure for patients with symptomatic NFPAs, with alow overall perioperative mortality rate similar to that inprevious reports. Cavernous sinus invasion was associatedwith STR, as safe complete resection of tumor lateral to thecarotid artery remains nearly impossible. Those patientswith STRs without adjuvant radiotherapy had higher ratesof recurrence than all other groups. Most important, GTR,with or without radiotherapy, resulted in rates of recur-rence-free survival similar to those in patients who receivedSTR with radiotherapy. Radiotherapy and/or STR appearedto be associated with higher mortality rates than thosefound in either the group of patients who underwent a GTRalone or the general population.

A broad range of results in the literature has compli-cated the interpretation of the role of radiotherapy in themanagement of NFPAs. In previous studies, patients whounderwent surgery alone had recurrence-free survival rang-ing from 48 to 94% at 5 years, and 14 and 56% at 10 years,depending on the institution.7,10,18,22,28,36 For patients whoalso have undergone radiotherapy, the 5-year progression-free survival rate has been reported to range from 93 to97%.5,15–17,25,28 The majority of previous studies did not ex-amine extent of resection as an important predictor of re-currence. When stratified by extent of resection, our datademonstrated that STR alone had similarly high rates of re-currence (~ 45% at 10 years). In contrast, GTR resulted infar lower rates of recurrence (11%), with no statistical dif-ference when followed by radiotherapy (9%).

Although increased mortality rates for patients withsome hypersecreting pituitary adenomas are well known,few studies have addressed survival after surgery specif-ically for NFPAs. Although the cause of death was not

E. F. Chang et al.


TABLE 5Recurrence-free survival estimates after surgery for NFPAs

5-yr 10-yr 15-yr 20-yr

Probability At Risk Probability At Risk Probability At Risk Probability At Risk

all (660 patients) 0.93 334 0.87 189 0.81 96 0.74 36GTR

no radiotherapy (248 patients) 0.96 103 0.91 61 0.86 25 0.79 8radiotherapy (60 patients) 0.96 41 0.91 27 0.87 16 0.75 7

STRno radiotherapy (75 patients) 0.72 26 0.55 17 0.45 3 0.45 1radiotherapy (277 patients) 0.96 164 0.89 84 0.85 52 0.81 20

FIG. 2. Kaplan–Meier plots of overall survival after resection ofNFPA. A: Total population (663 patients). B. Subset of patientswho survived . 10 years after surgery (484 patients). Patients whounderwent GTR without radiotherapy appeared to have improvedoverall survival compared with all other groups (*p , 0.05, log-rank test).

known for the majority of cases reviewed here, our studyconfirms the finding of increased delayed mortality for pa-tients with NFPAs (after 10 years following surgery), al-though only for those who undergo STR and/or radiationtherapy. Other studies have carefully described an asso-ciation with cardiovascular and cerebrovascular diseasepossibly as a result of hypopituitarism, radiotherapy, orboth3,12,23,35—although firm conclusions regarding this issueare difficult to make. A prospective study to assess hor-mone levels pre- and postoperatively, and also for manyyears after radiotherapy, will be helpful to definitively sortout this important issue.

These data support the limited application of radiothera-py for STR and the elimination of routine radiotherapy af-ter GTR. They also highlight the importance of GTR as theoperative goal. Greater hospital volumes and surgeon expe-rience have been clearly related to improved outcomes.2,9

Gross-total resection should be performed with preserva-tion of the normal gland. In some cases, patients with pre-operative hypopituitarism can exhibit recovery of pituitaryfunction after tumor debulking1—depending on the dura-tion of symptoms—thereby avoiding life-long hormonalreplacement.

Although the data here provide useful estimates of long-term recurrence and overall survival, direct comparisonbetween groups that did or did not receive radiotherapyshould take into consideration that the number per groupand some baseline tumor and patient characteristics weredifferent. Patients who underwent radiotherapy tended toharbor larger and more invasive tumors and had longer fol-low-up. Each of these factors probably introduced someselection bias to the assignment of radiotherapy and theability to identify recurrences, although our estimates forrecurrence should have been adjusted for some of theseconsiderations. A randomized trial would be the most idealstudy design to compare radiotherapy groups, but this isunlikely to happen given the lack of clinical equipoise, im-practically long follow-up, and large patient populationrequired to adequately power such a study. A more reason-able approach may be to address timing of radiotherapy af-ter STR for NFPAs, either immediately postoperatively orafter documented recurrence. Furthermore, modern man-agement of pituitary adenomas has evolved recently, withthe introduction of endonasal transsphenoidal techniques8,38

as well as stereotactic radiosurgery21,24,30,33 (although thesemodifications have not yet demonstrated superior out-comes).

Lastly, some authors have suggested that silent corti-cotroph adenomas, that is, NFPAs that stained positivelyfor ACTH but had no demonstrable hypercortisolemia, aremore aggressive and have higher recurrence rates than nullcell adenomas.31 Authors of an institutional series foundthat silent corticotroph adenomas had more invasive prop-erties overall but did not have increased recurrence rates.6In the 6% of patients in our study, however, there was noclear indication that those with silent corticotrophs hadhigher recurrence rates than patients with null cell adeno-mas, possibly because of the small number of patients. Animportant question for future studies will be to address howthe risk of recurrence of silent corticotrophs is impacted bythe extent of resection.

Conclusions

Meticulous microdissection with complete surgicalremoval is an effective and safe treatment for NFPAs, andshould be the goal for patients without cavernous sinusinvasion with attempts to preserve the compressed, remain-ing gland. Although complete resection of the tumor doesnot ensure cure in all cases, cure cannot occur unless com-plete resection of the tumor is the primary goal. Gross-total



743

TABLE 6Overall survival estimates after surgery for NFPAs

5-yr 10-yr 15-yr 20-yr

Probability At Risk Probability At Risk Probability At Risk Probability At Risk

all (660 patients) 0.91 603 0.81 434 0.69 295 0.55 127GTR

no radiotherapy (248 patients) 0.93 229 0.79 116 0.73 85 0.58 24radiotherapy (60 patients) 0.88 54 0.85 50 0.69 35 0.49 20

STRno radiotherapy (75 patients) 0.79 61 0.72 51 0.59 25 0.4 10radiotherapy (277 patients) 0.94 259 0.8 217 0.67 150 0.5 73

TABLE 7Predictors of death: uni- and multivariate Cox regression

Logistic Regression Analyses HR (95% CI) p Value

univariateage 1.09 (1.08–1.17) ,0.0001size .25 mm 1.37 (1.05–1.79) 0.02male sex 1.61 (1.25–2.08) 0.002extension

suprasellar 0.81 (0.62–1.05) 0.12infrasellar 0.87 (0.63–1.22) 0.42cavernous sinus invasion 1.19 (0.78–1.83) 0.41

pathology: null cell adenoma 1.52 (0.21–10.9) 0.67pathology: silent corticotroph 0.78 (0.08–7.40) 0.82STR 1.27 (1.01–1.62) 0.04radiotherapy 1.29 (1.00–1.67) 0.05

multivariate age 1.09 (1.08–1.84) ,0.0001GTR & no radiotherapy 0.72 (0.54–0.94) 0.02

resection leads to high control of long-term tumor growthwithout radiotherapy, and the use of routine postoperativeradiotherapy in this population did not provide significantclinical benefit and in fact was independently associatedwith lower long-term survival. In patients with STRs, ra-diotherapy can provide a greater recurrence-free survival.However, in all groups recurrence is relatively low and oc-curs after long durations and therefore must be weighedagainst the known complications of radiotherapy.

The optimum timing of radiotherapy remains unknownbecause of the uncertainty of the behavior of residual tumorand its potential for growth. Based on the current study,not all patients with STR of tumors experience progression,and progression is typically delayed. With improved im-aging modalities and the development of stereotactic ra-diosurgery, it may be possible to observe these patientswith serial imaging and defer radiotherapy until recurrenceis documented or use radiosurgery to treat the cavernous si-nus residual disease, assuming complete resection of thesellar and suprasellar portion of the tumor has been per-formed. In all patients, long-term follow-up is necessarygiven that the median time to recurrence is . 5 years.

References

1. Arafah BM, Kailani SH, Nekl KE, Gold RS, Selman WR: Imme-diate recovery of pituitary function after transsphenoidal resectionof pituitary macroadenomas. J Clin Endocrinol Metab 79:348–354, 1994

2. Barker FG II, Klibanski A, Swearingen B: Transsphenoidal sur-gery for pituitary tumors in the United States, 1996-2000: mortal-ity, morbidity, and the effects of hospital and surgeon volume. JClin Endocrinol Metab 88:4709–4719, 2003

3. Brada M, Ashley S, Ford D, Traish D, Burchell L, Rajan B: Cere-brovascular mortality in patients with pituitary adenoma. Clin En-docrinol (Oxf) 57:713–717, 2002

4. Brada M, Ford D, Ashley S, Bliss JM, Crowley S, Mason M, et al:Risk of second brain tumour after conservative surgery and radio-therapy for pituitary adenoma. BMJ 304:1343–1346, 1992

5. Brada M, Rajan B, Traish D, Ashley S, Holmes-Sellors PJ, NusseyS, et al: The long-term efficacy of conservative surgery and ra-diotherapy in the control of pituitary adenomas. Clin Endocrinol(Oxf) 38:571–578, 1993

6. Bradley KJ, Wass JA, Turner HE: Non-functioning pituitary ade-nomas with positive immunoreactivity for ACTH behave moreaggressively than ACTH immunonegative tumours but do not re-cur more frequently. Clin Endocrinol (Oxf) 58:59–64, 2003

7. Bradley KM, Adams CB, Potter CP, Wheeler DW, Anslow PJ,Burke CW: An audit of selected patients with non-functioning pi-tuitary adenoma treated by transsphenoidal surgery without irradi-ation. Clin Endocrinol (Oxf) 41:655–659, 1994

8. Cappabianca P, Cavallo LM, de Divitiis E: Endoscopic endonasaltranssphenoidal surgery. Neurosurgery 55:933–941, 2004

9. Ciric I, Ragin A, Baumgartner C, Pierce D: Complications oftranssphenoidal surgery: results of a national survey, review of theliterature, and personal experience. Neurosurgery 40:225–227,1997

10. Comtois R, Beauregard H, Somma M, Serri O, Aris-Jilwan N,Hardy J: The clinical and endocrine outcome to trans-sphenoid-al microsurgery of nonsecreting pituitary adenomas. Cancer 68:860–866, 1991

11. Cowper DC, Kubal JD, Maynard C, Hynes DM: A primer andcomparative review of major US mortality databases. Ann Epide-miol 12:462–468, 2002

12. Erfurth EM, Bulow B, Svahn-Tapper G, Norrving B, Odh K, Mik-

oczy Z, et al: Risk factors for cerebrovascular deaths in patientsoperated and irradiated for pituitary tumors. J Clin EndocrinolMetab 87:4892–4899, 2002

13. Ezzat S, Asa SL, Couldwell WT, Barr CE, Dodge WE, VanceML, et al: The prevalence of pituitary adenomas: a systematic re-view. Cancer 101:613–619, 2004

14. Fisher BJ, Gaspar LE, Noone B: Radiation therapy of pituitaryadenoma: delayed sequelae. Radiology 187:843–846, 1993

15. Flickinger JC, Nelson PB, Martinez AJ, Deutsch M, Taylor F: Ra-diotherapy of nonfunctional adenomas of the pituitary gland. Re-sults with long-term follow-up. Cancer 63:2409–2414, 1989

16. Gittoes NJ: Radiotherapy for non-functioning pituitary tumors—when and under what circumstances? Pituitary 6:103–108, 2003

17. Gittoes NJ, Bates AS, Tse W, Bullivant B, Sheppard MC, ClaytonRN, et al: Radiotherapy for non-function pituitary tumours. ClinEndocrinol (Oxf) 48:331–337, 1998

18. Greenman Y, Ouaknine G, Veshchev I, Reider G II, Segev Y,Stern N: Postoperative surveillance of clinically nonfunctioningpituitary macroadenomas: markers of tumour quiescence and re-growth. Clin Endocrinol (Oxf) 58:763–769, 2003

19. Hall WA, Luciano MG, Doppman JL, Patronas NJ, Oldfield EH:Pituitary magnetic resonance imaging in normal human volun-teers: occult adenomas in the general population. Ann InternMed 120:817–820, 1994

20. Hardy J: Transsphenoidal surgery of hypersecreting pituitary tu-mors, in Kohler PO, Ross GT (eds): Diagnosis and Treatmentof Pituitary Tumors. International Congress Series, Vol 303.Amsterdam: Excerpta Medica, 1973, pp 179–198

21. Kobayashi T, Mori Y, Uchiyama Y, Kida Y, Fujitani S: Long-term results of gamma knife surgery for growth hormone-produc-ing pituitary adenoma: is the disease difficult to cure? J Neuro-surg 102 (Suppl):119–123, 2005

22. Lillehei KO, Kirschman DL, Kleinschmidt-DeMasters BK, Ridg-way EC: Reassessment of the role of radiation therapy in the treat-ment of endocrine-inactive pituitary macroadenomas. Neurosur-gery 43:432–439, 1998

23. Lindholm J, Nielsen EH, Bjerre P, Christiansen JS, Hagen C, JuulS, et al: Hypopituitarism and mortality in pituitary adenoma. ClinEndocrinol (Oxf) 65:51–58, 2006

24. Losa M, Valle M, Mortini P, Franzin A, da Passano CF, CenzatoM, et al: Gamma knife surgery for treatment of residual nonfunc-tioning pituitary adenomas after surgical debulking. J Neurosurg100:438–444, 2004

25. McCollough WM, Marcus RB Jr, Rhoton AL Jr, Ballinger WE,Million RR: Long-term follow-up of radiotherapy for pituitaryadenoma: the absence of late recurrence after greater than or equalto 4500 cGy. Int J Radiat Oncol Biol Phys 21:607–614, 1991

26. Minniti G, Traish D, Ashley S, Gonsalves A, Brada M: Risk ofsecond brain tumor after conservative surgery and radiotherapyfor pituitary adenoma: update after an additional 10 years. J ClinEndocrinol Metab 90:800–804, 2005

27. Nammour GM, Ybarra J, Naheedy MH, Romeo JH, Aron DC:Incidental pituitary macroadenoma: a population-based study.Am J Med Sci 314:287–291, 1997

28. Park P, Chandler WF, Barkan AL, Orrego JJ, Cowan JA, GriffithKA, et al: The role of radiation therapy after surgical resection ofnonfunctional pituitary macroadenomas. Neurosurgery 55:100–107, 2004

29. Peace KA, Orme SM, Sebastian JP, Thompson AR, Barnes S,Ellis A, et al: The effect of treatment variables on mood and socialadjustment in adult patients with pituitary disease. Clin Endo-crinol (Oxf) 46:445–450, 1997

30. Picozzi P, Losa M, Mortini P, Valle MA, Franzin A, Attuati L, etal: Radiosurgery and the prevention of regrowth of incompletelyremoved nonfunctioning pituitary adenomas. J Neurosurg 102(Suppl):71–74, 2005

31. Scheithauer BW, Jaap AJ, Horvath E, Kovacs K, Lloyd RV,Meyer FB, et al: Clinically silent corticotroph tumors of the pitu-itary gland. Neurosurgery 47:723–730, 2000

E. F. Chang et al.




745

32. Schisterman EF, Whitcomb BW: Use of the Social SecurityAdministration Death Master File for ascertainment of mortalitystatus. Popul Health Metr 2:2, 2004

33. Sheehan JP, Niranjan A, Sheehan JM, Jane JA Jr, Laws ER,Kondziolka D, et al: Stereotactic radiosurgery for pituitary ade-nomas: an intermediate review of its safety, efficacy, and role inthe neurosurgical treatment armamentarium. J Neurosurg 102:678–691, 2005

34. Snyder PJ, Fowble BF, Schatz NJ, Savino PJ, Gennarelli TA:Hypopituitarism following radiation therapy of pituitary adeno-mas. Am J Med 81:457–462, 1986

35. Tomlinson JW, Holden N, Hills RK, Wheatley K, Clayton RN,Bates AS, et al: Association between premature mortality and hy-popituitarism. West Midlands Prospective Hypopituitary StudyGroup. Lancet 357:425–431, 2001

36. Turner HE, Stratton IM, Byrne JV, Adams CB, Wass JA: Audit ofselected patients with nonfunctioning pituitary adenomas treated

without irradiation—a follow-up study. Clin Endocrinol (Oxf)51:281–284, 1999

37. Wilson CG: Neurosurgical management of large and invasivepituitary tumors, in Tindall GT, Collins WF (eds): Clinical Man-agement of Pituitary Disorders. New York: Raven Press, 1979,pp 335–342

38. Zada G, Kelly DF, Cohan P, Wang C, Swerdloff R: Endonasaltranssphenoidal approach for pituitary adenomas and other sellarlesions: an assessment of efficacy, safety, and patient impressions.J Neurosurg 98:350–358, 2003

Manuscript submitted May 14, 2007. Accepted July 30, 2007. Address correspondence to: Edward F. Chang, M.D., Department

of Neurological Surgery, University of California, San Francisco,505 Parnassus Avenue, M779 Campus Box 0122, San Francisco,California 94143. email: [email protected].

Long-term recurrence and mortality after surgery and adjuvant radiotherapy for nonfunctional pituitary adenomas

Documents