Decline of posttreatment tumor marker levels ... - Wiley Online Library

1586

Decline of Posttreatment Tumor Marker Levels after Therapy of Nonsmall Cell Lung Cancer A Useful Outcome Predictor C. Harris Spiridonidis, M.D.," Leslie R. Laufman, M.D.,* Kathy A . Sfydnicki, R.N.,* Judith W . Noltimier, M.D.,* Chuck C. Cho, M.D.," Donn C. Young, Ph.D.,t WilliamJ.Hicks, M.D.," Mark L. Segal, M.D.,*Jery T . Guy, M.D.,* and Bernard L. Zidar, M.D.S

Background. The assessment of treatment efficacy in nonsmall cell lung cancer (NSCLC) is limited by the lack of a clear association between clinical response and survival. The prognostic usefulness of treatment-induced tumor-marker declines in NSCLC has not been established. The authors investigated the prognostic significance of treatment-induced declination in tumor marker levels of carcinoembryonic antigen, CA 19-9,and CA 125 in a group of patients with NSCLC treated with a brief course of cisplatin-based chemotherapy. Methods. Eighty-three patients with NSCLC enrolled on 2 related treatment protocols had pretreatment tumormarker determinations. Patients were restaged 10 to 12 weeks after study entry, and clinical and marker responses were determined. Results. Thirty-eight patients (46%) had elevated pretreatment tumor markers, 36 (42%) of whom were evaluable for both clinical and marker responses. Pretreatment, the latter 36 individuals had measurable or evaluable disease, and at least one elevated tumor marker (greater than twice normal); posttreatment, they had follow-up measurements of both parameters. Of the 36 patients, 8 had normalization of tumor marker levels, 13 had 50-99% marker level declination, and 15 had less than 50% or no declination. In the same group of 36 patients, there were, 1 patient with complete clinical response, 11 with partial response, 19 with stable disease, From *The Columbus Community Clinical Oncology Program (CCOP), Columbus, Ohio; tThe Ohio State University Comprehensive Cancer Center, Columbus, Ohio; and $The Allegheny Community Clinical Oncology Program (CCOP), Pittsburgh, Pennsylvania. The authors thank Patricia Dunn, B.S., R.N., for excellent data coordination. Address for reprints: C. Harris Spiridonidis, M.D., 8100 Ravines Edge Court, Columbus, OH 43235. Received June 22, 1994; revision received September 19, 1994, and November 28,1994; accepted December 7,1994.

and 5 with progressive disease. Marker responses occurred with equal frequency in clinical responders and nonresponders. There was no association between clinical response and survival, but there was a strong association between marker response and survival. Conclusions. In patients with nonsmall cell lung cancer with elevated pretreatment tumor marker levels, treatment-induced marker level declination can be a surrogate indicator for survival. Cancer 1995;75:1586-93. Key words: tumor marker levels, lung cancer, marker response, survival prediction.

Advanced nonsmall cell lung cancer (NSCLC)is associated with a poor prognosis. Intensive platinum based chemotherapy in metastatic disease, the use of combined modality treatment for locally advanced tumors, and the introduction of new active agents have had only a limited impact on the natural history of this malignancy.1-5 Evaluation of new antineoplastic agents and strategies have been hindered in many studies by the inability to demonstrate an association between clinical response and survival time. In a prospective single treatment arm trial, 71 patients with advanced NSCLC were treated with a brief, high dose cisplatin chemotherapy regimen. In another related trial, 15 patients with locally advanced disease were treated with concurrent thoracic irradiation and cisplatin based chemotherapy. In 83 of 86 patients, tumor marker levels were determined at study entry; marker levels were measured again after treatment completion 10-12 weeks later. Clinical response rates and survival times of these patients were comparable to those reported in other recent trials. However, because clinical evaluations and

NSCLC Marker Level Declination and SurvivallSpiridonidis et al. marker level measurements were repeated at a predetermined time after treatment completion, it was possible to examine the relationships of both marker level response and clinical response with survival time. Patients and Methods

Patient Selection Criteria for study entry consisted of histologically confirmed NSCLC with inoperable intrathoracic disease, extrathoracic metastases, or recurrent disease after surgery or thoracic irradiation. All pathology specimens were reviewed by one of us (J.N.) for confirmation of diagnosis and histologic subtype. Requirements for study participation were a Zubrod performance status of 0-2; normal liver, renal, and cardiac function; no previous malignancy; and no prior chemotherapy. All patients were informed of the investigational nature of these protocols and gave consent to participate, in accordance with the United States Food and Drug Administration and institutional guidelines after approval by the institutional review boards in all participating institutions.

Pa tien t Population Between May 1983 and March 1988,71 eligible patients were entered in the chemotherapy study (VEP-I). Of these, 25 had newly diagnosed metastatic NSCLC, 21 had recurrent disease after surgery or irradiation, and 25 had intrathoracic inoperable disease. Between March 1988 and July 1989, 15 eligible patients with locally advanced NSCLC, Stages IIIA and IIIB, were entered in the combined modality study (VEP-2);all of these patients had newly diagnosed disease and had received no prior therapy. The characteristics of all eligible patients are listed in Table 1.

Treatment Schedule The 71 patients on the VEP-1 protocol received chemotherapy with cisplatin, etoposide, and vinblastine. Good risk patients, defined as 60 years of age or younger with performance status of 0-1 and no prior irradiation, were treated intravenously with vinblastine 5 mg/m2 intravenous bolus on day 1, etoposide 40 mg/ m2 daily infused during a period of 1-2 hours on days 1-5, and cisplatin 30 mg/m2 daily infused during a period of l hour on days 1-5. Poor risk patients were treated with reduced doses of vinblastine at 3 mg/m2 on day 1and etoposide at 30 mg/m2 on days 1-5. Cisplatin doses remained the same. Cycles were repeated every 3 weeks or upon recovery of absolute neutrophil counts

1587

Table 1. Patient Characteristics Eligible In = 861 Gender Male Female Zubrod performance status 0 1 2 Age (yr) Median Range Histology Adenocarcinoma Squamous Large cell Unclassified Measurable versus evaluable Measurable Evaluable Disease status 1/11

IIIA IIIB Metastatic Recurrent post radiation therapy Recurrent uost surgerv

With elevated markers (n = 381

56 (65) 30 ( 3 5 ) 9 (10) 60 (70) 17 (20)

58.5 39-73 41 (48) 28 (33) 15 (17) 2 (2)

49 (57) 37 (43) 5 (6) 20 (23) 15 (17) 25 (29) 19 (22) 2 (2)

Values in uarentheses are uercentares

to 1500/p1 and platelet counts to 75,OOO/pl with a maximum allowed delay of 2 weeks. Three cycles were administered. Patients were evaluated for response 4-6 weeks after the third chemotherapy cycle, i.e., 10-12 weeks after initiation of therapy. All tests initially positive for disease were repeated and biopsies performed when feasible. Patients with response to therapy and those who had stable disease did not receive additional treatment for at least 10 weeks. After that, any additional treatment was given according to the preferences of patients and their attending physicians, who generally were unaware of the tumor marker level data or their potential prognostic significance. Of 25 patients with locally advanced disease on the VEP-1 protocol, 21 eventually received thoracic irradiation, as did 12 of the 25 patients with metastatic disease and 1 of the 21 with recurrent disease. All 15 patients on the VEP-2 protocol were treated with concurrent chemotherapy and thoracic irradiation. Chemotherapy consisted of intravenous vinblastine 1 mg/m2 on days 1-3, etoposide 40 mg/m2 on days 1-3, and cisplatin 30 mg/m2 on days 1-3 of the first week

1588

CANCER April 2,2995, Volume 75, No. 7

of radiation. The day 1 doses were administered once weekly for each subsequent week of radiation treatments. Radiation was started on day 1and administered at a rate of 200 cGy daily, 5 days per week, for a total dose of 5800-6000 cGy during a period of 6 weeks. In patients with Grade 111-IV toxicity, both chemotherapy and irradiation were withheld until resolution of symptoms and were restarted concurrently at full dose. Patients were restaged approximately 4 weeks after completion of chemoradiotherapy, i e., as with the VEP-1 protocol, 10-12 weeks after study entry. Tumor marker levels were drawn in both patient groups on study entry and 10-12 weeks later. Pretreatment carcinoembryonic antigen (CEA), CA 19-9, and CA 125 values were considered elevated if their concentrations were equal to or greater than 10 pg/ml, 70 U/ml, and 70 U/ml, respectively; these values represent twice normal limits.

Definitions of Response Standard clinical response criteria were used. Complete clinical response (cCR)was defined as disappearance of all detectable tumor by imaging studies and clinical examination. Partial response (cPR)was defined as at least 50% reduction in the sum of products of the perpendicular diameters of measurable lesions, lasting for a minimum of 4 weeks. Stable disease (cSD) was defined as tumor regression of less than 50% or tumor growth of less than 25 YO compared with pretreatment evaluation. Progressive disease (cPD) was defined as a more than 25% increase in the area of existing tumors or the appearance of new lesions. For evaluable lesions, cPR was defined as an estimated decrease in the tumor size of at least 50%. Marker level responses were determined only in patients having at least one elevated pretreatment marker level. To determine response, pretreatment marker level values were compared with those obtained at restaging. Marker level normalization (mCR)was defined as a return to normal concentrations of all elevated marker levels (CEA to 5 pg/ml or less, CA 125 to 35 U/ ml or less, CA 19-9 to 35 U/ml or less). Marker level partial response (mPR) was defined as a decline in the concentration of each elevated marker level by at least 50% but not normalization. Any lesser declines and any increases were considered marker level nonresponses (mNR).

Statistics Statistical analyses of frequency data were performed using chi-square tests, McNemar’s modification of chisquare, or Fisher’s exact tests.6 Survival times were de-

termined from the date of study entry and from the date of restaging, 12 weeks after study entry, to the date of last follow-up or death. In univariate analyses of survival times, the Lee-Desu statistics program7 was used for comparisons of subgroups with actuarial survival data.8 Cox proportional hazards regression models’ with prognostic factors as covariates were used to analyze survival times in a multivariate setting. Results

Eligibility and Evaluability Through the statistical office of the Columbus Community Clinical Oncology Program (CCOP), 92 patients were registered and treated at institutions affiliated with either the Columbus or the Allegheny CCOP. As a result of pathology review (J.N.),six patients were found to be ineligible because the histologic type of their tumors was not consistent with NSCLC. All 86 eligible patients are included in the survival time calculations. No patients were lost to follow-up.

Tumor Marker Levels Tumor marker levels were determined in 83 of the 86 eligible patients before treatment initiation. Assays for CEA were available in all institutions throughout the study, and pretreatment CEA levels were obtained in 80 patients. Six patients lacked pretreatment CEA level data because of clerical errors (i.e., level measurements ordered but not drawn). Because of the unavailability of assays, CA 125 and CA 19-9 levels could not be determined in the 27 patients enrolled in the VEP-1 study between 1983 and 1985. Assays for these tumor marker levels became available in 1986; thus, CA 125 and CA 1-9 levels were obtained in 10 of 24 patients enrolled during that year. Finally, 33 of the 35 patients entered after 1986 had all three marker levels determined; in the remaining 2 patients, the marker level data were missing because of clerical errors. The range of CEA values was 0.2-1528 pg/ml, with a median of 5.9 &ml; CA 125 values ranged from 2 U/ml to 1820 U/ml, with a median of 24.3 U/ml; and CA 19-9 values ranged from 1.6 U/ml to 19,250 U/ml, with a median of 21 U/ml. Forty-five of the 83 patients had nonelevated pretreatment tumor marker levels. Twenty-five patients were classified on the basis of all three marker levels and 20 patients on the basis of CEA alone. The median survival time for this group of patients was 9.0 months. Of these 45 patients, 8 could not be evaluated for re-

NSCLC Marker Level Declination and Survival/Spiridonidis et al.

1589

Table 2. Pretreatment Tumor Marker Results Normal

Borderline

Elevated

70 11

CEA (ng/ml) No. of patients CA 125 (U/ml) No. of patients CA 19-9 (U/ml) No. of patients

No. of patients tested 80 44 43

CEA: carcinoembryonic antigen.

sponse: 5 died early during treatment, and 3 received inadequate chemotherapy trials (2 because of treatment refusal and 1 because of myocardial infarction). The clinical response rate (cCR cPR) in the 37 evaluable patients with nonelevated tumor marker levels was 37%. None of the 37 patients were found to have elevated tumor marker levels on restaging. One or more elevated tumor marker levels were found in 38 of the 83 (46%)patients before the initiation of treatment; two elevated marker levels were found in 9 patients, and three elevated marker levels in 5 patients. Pretreatment CEA values were elevated in 31 of 80 (39%) individuals tested for that marker level, CA 125 values were elevated in 15 of 44 (34%), and CA 199 values were elevated in 11 of 43 (26%) (Table 2). Pretreatment marker levels were elevated in 17 of the 35 (48%) patients with Stage 111 NSCLC, in 13 of the 25 (52%)patients with Stage IV disease, and in 8 of the 21 patients with recurrent cancer after surgery or radiation. In patients with Stage IV NSCLC, liver or bone involvement was present in 70% of those with elevated pretreatment marker levels and in 75% of those with no pretreatment marker level elevation. Thus, no clear association between marker level elevation and disease stage or specific organ involvement was evident. The pretreatment characteristics of the 38 patients with elevated tumor marker levels are listed in Table 1. Two of the 38 patients with elevated pretreatment tumor marker levels died early during their treatment, leaving 36 patients evaluable for clinical response and marker level response. All 36 evaluable patients received an adequate trial of therapy and had clinical and marker level evaluation pretreatment and posttreatment. One of the 36 did not complete the whole course of treatment because, while receiving therapy, he had disease progression by clinical and marker level criteria (CEA increased from 12.5 pg/ml to 24 pg/ml); however, this patient was included in all analyses. The derivation of the 36 patients evaluable for clinical and marker level response is shown in Table 3.

+

Clinical Responses Clinical response assessment revealed that among the 36 evaluable patients with elevated pretreatment tumor marker levels, 24 patients had no response (19 cSD and 5 cPD), and 12 had a response (1 cCR and 11 cPR), for a response rate of 33%. This did not differ significantly from the response rate of 37% in the clinically evaluable patients who had nonelevated pretreatment marker levels.

Tumor Marker Level Responses Marker level response assessment in the 36 evaluable patients with elevated pretreatment tumor marker levels showed 8 mCR, 13 mPR, and 15 mNR. Marker level responses occurred in 58% of patients with clinical response and those with no clinical response (Table 4). Concordance of clinical response with marker level response was observed in only 17 of 36 (47%) patients, and the two evaluations differed significantly by McNemar’s test (P = 0.039), indicating no association between clinical and marker level responses.

Interim and Relapse Tumor Marker Level Values Interim tumor marker level values, drawn before each of the three chemotherapy cycles, were available in 18

Table 3. Derivation of 36 Evaluable Patients Registered Wrong histology Eligible

92 -6 86

No pretreatment tumor markers Eligible with tumor markers

-3 -

Tumor markers not elevated Eligible with elevated tumor markers

-45 -

Early death Clinically evaluable and marker evaluable

83 38 -2

_ .

36

1590

CANCER April 1,1995, Volume 75, No. 7

Table 4. Marker Responses Versus Clinical Responses

'.Oi

n

Clinical response cCR Marker response mCR + mPR mNR Total

+ cPR

cSD + cPD

Total

14 10 24

21 15 36

e 7 5 12

61

cCR complete response, clinical; cPR: partial response, clinical; cSD: stable disease, clinical; cPD: progressive disease, clinical; mCR: complete response, marker: mPR: Dartial resDonse. marker: mNR: no resDonse. marker. 2;

patients, all on the chemotherapy (VEP- 1) study. Of these 18 patients, 7 had marker level response, and in all 7 the marker levels declined with each successive chemotherapy cycle. The maximum rate of marker level declination was observed in two of the seven patients after the first cycle, in three patients after the second cycle, and in two patients after the third cycle. From these limited data, it appears that marker level declines in patients with response were continuous throughout the 3 cycles of chemotherapy. Eleven patients with available interim marker levels had no marker level response. In four of these patients, marker levels declined by 17-4970 after the first cycle of chemotherapy but increased during subsequent chemotherapy to levels equal to, or greater than, the pretreatment values; marker levels increased in two patients throughout the 3 chemotherapy cycles; continuous declines that did not reach 50% occurred in three patients; in the remaining two patients, no discernible pattern was evident. Eleven patients with elevated pretreatment marker level values and marker level responses had marker levels checked again at the time of disease progression. All marker levels had increased before or concurrent with clinical progression, and in eight patients, relapse levels exceeded pretreatment levels. However, of the 12 patients with nonelevated pretreatment marker levels who had marker levels determined at the time of disease progression, only 2 had elevated levels.

'+!j7

,

.

C l i n I Z

Non-Rsrponders

00

0

24

12

36

Survival in Months

Figure 1. Survival curves for 36 evaluable patients with elevated pretreatment tumor marker levels according to clinical response. Survival is measured from study entry. (- - -): clinical responders (completeresponse, clinical partial response, clinical); (-): clinical nonresponders (stable disease, clinical progressivedisease, clinical).

+

+

+

For the 12 patients with clinical response (cCR cPR), the median survival time was 8.0 months, compared with 12 months for the 24 patients with no clinical response (cSD cPD; P = 0.32) Fig. 1). For the 21 patients with tumor marker level response (mCR mPR), the median survival time was 13.8 months, compared with 6.2 months for the 15 patients with no marker level response (mNR; P = 0.0002) (Fig. 2). There were no changes in P values when survival times were measured from the "landmark" time

+

+

Survival Times The median survival times for all 86 eligible patients was 8.5 months. Overall, 19 (22%) patients survived longer than 1 year, and 6 (7%) patients survived longer than 2 years. All 38 patients with elevated tumor marker levels had died at the time of analysis. Their median survival time was 7.7 months, which was not significantly different from the median survival time of 9.0 months in the 45 patients with nonelevated marker levels.

.- -

NOn-RespOndnr

00 0

24

12

36

Survival In Months

Figure 2. Survival curves for 36 evaluable patients with elevated pretreatment tumor marker levels according to marker level response. Survival is measured from study entry. (-): marker level responders (mCR mPR); (- - -): marker level nonresponders (mNR).

+

NSCLC Marker Level Declination and Survival/Spiridonidis et al.

71 I ' I

Survival in Months

Figure 3. Survival curves for 36 evaluable patients with elevated pretreatment tumor marker levels according to clinical response. Survival time is measured from the time of response evaluation, 12 weeks after study entry. (- - -): clinical responders (cCR cPR); (-): clinical nonresponders (cSD + cPD).

+

of 12 weeks after study entry (Figs. 3 and 4). All eight patients with mCR (clinical status: 1 cCR, 2 cPR, 4 cSD, 1 cPD) survived longer than 1 2 months. Discussion

The choice of endpoints to document antineoplastic activity of systemic therapy is problematic in NSCLC because clinical responses do not always correlate with improved survival times.l o The physical examination findings and imaging studies traditionally used to assess tumor response are not pure measurements of tumor burden; rather, they are indirect approximations of tumor mass that include normal and malignant tissue. Thus, the poor correlation between radiologic and pathologic response in NSCLC is not surprising." In addition, interobserver variability in response evaluations is significant," and only half of all patients with metastatic NSCLC have bidimensionally measurable lesions. The classificationof patients with stable disease as those with no response also is arguable because such patients are a heterogeneous group that includes patients with refractory but inherently slow growing tumors and patients whose tumors have responded to treatment but whose imaging studies show residual scarring or edema. In either case, the outcome for these patients is likely to be different from that of patients with resistant, rapidly growing tumors. The poor correlation between clinical response and survival times reported in several recent NSCLC t r i a l ~ ' ~and - ' ~ the lack of a survival benefit for patients with response when

1591

compared with that of patients with stable d i ~ e a s e ' ~ , ' ~ can be explained on the basis of these considerations. To overcome the limitations of clinical response assessments, tumor marker level declines have been used as outcome predictors in multiple myeloma and in germ cell and ovarian cancers to supplement clinical evaluations. More recently, prostate-specific antigen tumor marker level declines have been shown to be the most useful surrogate endpoints for chemotherapy effectiveness in hormonally refractory prostatic cancer.18,19 In NSCLC there is no universally elevated tumor marker level, and the prognostic value of posttherapy marker level declines has not been investigated previously. In this study, the use of CEA, CA 19-9, and CA 125 tumor marker levels allowed marker level response evaluation in 42% of patients. This proportion might have been higher if levels for the three marker levels had been obtained in all eligible patients. Indeed, one of the study limitations was the lack of information regarding pretreatment CA 125 and CA 19-9 values in approximately half of the eligible patient population. However, the determining factor for missing CA 125 and CA 19-9 data was the unavailability of assays for these marker levels at the enrolling institution on study entry and not the disease stage or prognosis; thus, we do not believe that missing marker level data reflected systemic biases. A potential problem in interpreting our results is that a proportion of the 20 patients, considered to belong in the nonelevated marker level group on the basis of a nonelevated CEA alone, might have had elevations

2 1

I

' 1 Marker Response

00

1 0

Responder

,

, 12

I

,

j,- -&-Responder

24

Survival in Months

Figure 4. Survival curves for 36 evaluable patients with elevated pretreatment tumor marker levels according to marker level response. Survival time is measured from the time of response evaluation, 12 weeks after study entry. (-): marker level responders (mCR + mPR); (- - -): marker level nonresponders (mNR).

1592

CANCER April 2,1995,Volume 75, No. 7

of the missing marker levels CA 125 or CA 19-9 and thus may have been misclassified. We analyzed 29 patients with nonelevated pretreatment CEA levels and available values for the other two marker levels; 25 (86%) of these individuals also had nonelevated CA 125 and CA 19-9 levels. Thus, we believe that we correctly classified most of the 20 patients on the basis of nonelevated CEA levels alone and that any potential misclassification in this regard affected only a small number of individuals. Pretreatment tumor marker levels did not correlate with clinical response or survival time. However, treatment induced marker level decline of 50% or more predicted for longer survival. This strong association between marker level response and survival time occurred in a patient group in which clinical response did not correlate with survival time. Previous attempts to correlate clinical responses with survival times have been justifiably criticized for the exclusion of patients with early demise or early progression and the use of "best" response as a parameter, regardless of the interval between initiation of therapy and the development of that response.*' Both practices confound the measurement of survival time and skew attempts to correlate response and survival time. In this study, such problems with analysis did not occur because the patients were evaluated for marker level and clinical response at two fixed landmark times: namely, immediately before they started the brief course of cisplatin based therapy and 10-12 weeks after initiation of that intervention. Increased survival time in patients with clinical response often is an artifact of the duration of treatment and not the result of such treatment. This problem previously has been approached by using time under treatment or time from initiation of treatment to response as covariates or by measuring survival time from the end of treatment, rather than from treatment entry. In this study, the duration of treatment was uniform (approximately 8 weeks in the VEP-1 and VEP-2 treatment arms), and response determinations were made at the same time for all patients. Because most of the patients with elevated tumor marker levels (36 of 38) had survived to this landmark, the impact of pretreatment prognostic factors and of the treatment itself on shortterm survival to that time was minimized. Thus, because there was no association between treatment duration and response, the need for covariate adjustment was obviated. Responses are both outcomes and predictors. Measuring survival times from the end of treatment, rather than from study entry, can convert a response to a baseline factor provided that patients with no response do not terminate treatment early. In this study, all 36 eval-

uable patients had by definition survived to the time of response evaluation, and 14 of the 15 patients with mNR had received all treatments as planned before restaging for marker level response. Thus, measuring survival times from restagng, rather than from study entry, did not change our results. A valid criticism of our marker level response definition is that it does not provide for the possibility of an initially nonelevated marker level value becoming elevated at restaging in a patients with a marker level response. Six of our patients with marker level response had one elevated marker level and one or two normal marker levels before treatment. In none of the six was a previously normal marker level found to be elevated at restaging, but we recognize that this may occur with larger patient numbers. Another question relates to the possibility that declines in CEA values predict for survival better than does a combination of the three marker levels. We analyzed CEA values alone (data not shown) and found that declines equal to or greater than 50% were associated with improved survival times ( P = 0.012), but the association was not as strong as when values for all three marker levels were used ( P = 0.0002). A potential limitation of our study is that because of unrecognized biases, our relatively small sample of patients with elevated marker levels may not be representative of the larger NSCLC population; however, the overall outcome in this group, as measured by clinical response rates (33% versus 39%) and median survival times (7.7 months versus 8.5 months), did not differ significantly from that of the whole eligible patient population. Larger patient numbers, with determinations of CEA, CA 125, and CA 19-9 in all patients, are needed for a more accurate estimate of the proportion of patients with NSCLC and elevated marker levels, for direct systemic analysis of the relative importance of individual marker level responses, and for confirmation of our results obtained in this small cohort of patients. In conclusion, treatment induced marker level declination correlated closely with improved survival times in patients with NSCLC. Unlike weight and performance status, tumor marker level determinations are not affected by short-term treatment toxicities. Confirmation of these preliminary findings in a larger patient population is likely to result in more widespread use of serial tumor marker level measurements to assess treatment effectiveness in advanced NSCLC. References 1. Vokes EE, Vijayakumar S, Bitran JD, Hoffman PC, Golomb HM. Role of systemic therapy in advanced non-small-cell lung cancer. Am I M e d 1990;89:777-86.

NSC LC Marker Level Declination and Survival/Spiridonidis et al. 2. Rapp E, Pater JL, Willan A, Cormier Y, Murray N, Evans WK, et al. Chemotherapy can prolong survival in patients with advanced non-small-cell lung cancer: report of a Canadian multicenter randomized trial. J Clin Oncol 1988; 6:633-41. 3. Weiden PL, Piantadosi S. Preoperative chemotherapy (cisplatin and fluorouracil) and radiation therapy in Stage I11 non-smallcell lung cancer: a Phase I1 study of the Lung Cancer Study Group. J Natl Cancer lnst 1991;83:266-72. 4. Dillman RO, Seagren SL, Propert KJ, Guerra J, Eaton WL, Perry MC, et al. A randomized trial of induction chemotherapy plus high-dose radiation versus radiation alone in Stage 111 nonsmall-cell lung cancer. N Engl IMed 1990;323:940-5. 5. Murphy WK, Fossella FV, Winn RJ, Dong MS, Hynes HE, Gross HM, et al. Phase 11 study of taxol in patients with untreated advanced non-small-cell lung cancer. J Natl Cancer rnst 1993;85: 384-8. 6. Sokal RR, Rohl FJ. Biometry. San Francisco: WH Freeman and Company, 1969. 7. Lee E, Desu M. A computer program for comparing K samples with right-censoyed data. Comp Prog Biomed 1972;3:315. 8. Berkson J, Cage R. Calculation of survival rates for cancer. Proc Mayo Clinic 1950;25:270. 9. Cox DR. Regression models and life tables. J R Stat Soc, Ser B 1992; 34:187. 10. Livingston RB. Stage IV non-small-cell lung cancer: the guides are perplexed. ] Clin Oncol 1989; 7:1591-3. 11. Martini N, Kris MG, Gralla RJ, Bains MS, McCormack PM, Kaiser LR, et al. The effects of preoperative chemotherapy on the resectability of non-small cell lung cancer with mediastinal lymph node metastases (N2 MO). Ann Thorac Surg 1988;45:370-9. 12. Herschorn S, Hanley J, Wolkove N, Cohen C, Frank H, Palayew

13.

14.

15.

16.

17.

18.

19.

20.

1593

M, et al. Measurability of non-small-cell lung cancer on chest radiograms. J Clin Oncol 1986;4:1184-90. Luedke DW, Einhorn LH, Omura GA, Sarma PR, Bartolucci AA, Birch R, et al. Randomized comparison of two combination regimens versus minimal chemotherapy in non-small-cell lung cancer: a Southeastern Cancer Study Group trial. J Clin Oncol 1990;8:886-91. Bonomi PD, Finkelstein DM, Ruckdeschel JC, Blum RH, Green MD, Mason B, et al. Combination chemotherapy versus single agents followed by combination chemotherapy in Stage IV nonsmall-cell lung cancer. 1Clin Oncol 1989; 7:1602-13. Weick JK, Crowley J, Natale RB, Hom BL, Rivkin S, Coltman CA, et al. A randomized trial of five cisplatin-containing treatments in patients with metastatic non-small-cell lung cancer: a Southwest Oncology Group study. ] Clin Oncol 1992;9:1157-62. Cellerino R, Tummarello D, Guidi F, Isidori P, Raspugli M, Biscottini 8, et al. A randomized trial of alternating chemotherapy versus best supportive care in advanced non-small-cell lung cancer. JClin Oncol 1991;9:1453-61. Ecklund E, Miller ME, Ansari R, Fisher WB, Einhom LH. Phase I1 trial of high-dose cisplatin plus etoposide plus vinblastine in non-small-cell lung cancer: a Hoosier Oncology Group Study. A m J C l i n Oncol 1991;14:412-5. Myers C, Cooper M, Stein C, LaRocca R, Walther MM, Weiss G, et al. Suramin: a novel growth factor antagonist with activity in hormone-refractory metastatic prostate cancer. J Clin Oncol 1992; 10:881-9. Kelly WK, Scher HI, Mazumdar M, Vlamis V, Schwartz M, Fossa SD. Prostate-specific antigen as a measure of disease outcome in metastatic hormone-refractory prostate cancer. J Clin Oncol 1993; 11:607-15. Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. ] Clin Oncol 1983; 1:710-9.