Docstoc

Regional variation in breast-conserving surgery and - The National

Document Sample
Regional variation in breast-conserving surgery and - The National Powered By Docstoc
					     Regional variation in
breast-conserving surgery and
radiotherapy for breast cancer


   Paul M. Walsh
           Background and objectives
• Randomized trials of breast cancer treatment show survival
  equivalent for breast-conserving surgery (BCS) + radiotherapy
  v. mastectomy.
• Use of BCS has increased over time in ROI
• Survival and adjuvant therapy for breast cancer varies
  geographically within ROI
• How much does use of BCS vary? – and why?
• Is it appropriately combined with radiotherapy, i.e. adequate
  breast-conserving therapy (BCT)?
• Implications for recurrence/survival?
    2000 Breast Cancer Management
 Clinical Guidelines (Royal College of Surgeons in Ireland)
• “All patients should be considered, but not all are suitable, for BCS.”
• “Appropriate conservative surgery to the breast followed by
  radiotherapy gives similar survival to more radical procedures.”
• “Contraindications:
    multifocal disease
    2 or more primary tumours in different quadrants
    anticipated poor cosmetic result
    1st & 2nd trimester of pregnancy
    inability to obtain histologically clear margins
    patient preference
    contraindications to radiotherapy.”
• “Tumour size per se is not a contraindication for BCS.”
          1990 NIH Consensus Conference
 Appropriate surgical treatment for Stage I or II breast
   cancer is:
 • mastectomy (total breast-removal) or
 • breast-conserving surgery plus radiotherapy

1998 Canadian Association of Radiation Oncologists:
     Steering Cttee on breast cancer treatment
• “Women who undergo BCS should be advised to have
  postoperative breast irradiation.”
• “Omission of radiotherapy after BCS almost always
  increases the risk of recurrence”
  Appropriate use of BCS plus radiation
         therapy (NCI PDQ website)
• “Breast-conserving surgery alone without radiation therapy has been
  compared with breast-conserving surgery followed by radiation
  therapy in six prospective randomized trials ... Every trial
  demonstrated a lower in-breast recurrence rate with radiation
  therapy, and this effect was present in all patient subgroups.”
• “Although all trials ... have shown highly statistically significant
  reductions in local recurrence rate, no single trial has demonstrated
  a statistically significant reduction in mortality. However, in the
  2005 Early Breast Cancer Trialists' Collaborative Group's (EBCTCG)
  update, when all relevant trials were combined, 15-year breast
  cancer mortality was reduced from 35.9% to 30.5% in women
  receiving radiation therapy ... similar effect on all-cause mortality.”
          Methods: study population
• Women age 15-99, resident in ROI, diagnosed with
  invasive breast cancer.
• DCOs, autopsy-only and later malignancies excluded.
• BCS v mastectomy: 8592 patients diagnosed 2000-2004
  who had either mastectomy (total breast-removal) or BCS
  within 6 months of diagnosis.
• BCS +/- radiotherapy: 4019 patients who had BCS as their
  main surgical procedure (excluding those who had a later mastectomy
  within 12 months; some analyses exclude patients surviving <12 mths)
• Focus on patients with local or regional disease, and
  assumption that radiotherapy is adjuvant.
        Methods: treatment models
• Relative risks derived by Poisson regression with
  robust error variance*, equivalent to logistic
  regression but with output expressed as risk
  ratios rather than odds ratios.
• Models selected by likelihood ratio comparison
  of equivalent logistic models (LR test in Stata may
  not be appropriate for GLMs with robust error variance).


  *Zou G. A modified Poisson regression approach to prospective studies
  with binary data. Am J Epidemiol 2004; 159(7):702-6.
         Methods: relative survival
• Relative survival (compared with general
  population) assessed using strs algorithm in Stata
  (Dickman et al.) – 2000-04 diagnoses with
  follow up to end of 2005 [To be extended]
• Excess hazards (associated with breast cancer)
  modelled using a GLM with Poisson error
  structure in Stata (Dickman et al. 2004)
  – excess hazard ratios or relative excess risk

  *Dickman PW et al. Regression models for relative survival. Statist
  Med 2004; 23: 51-64.
   Regional variation in breast cancer
treatment: area of residence (1995-99 v 2000-04)




                         chemotherapy %
 surgery %




                         hormone therapy %
 radiotherapy %
      Regional variation in % BCS use
                    (2000-04, all ages & stages)
 47%
overall




                    59%                                61%
      32%                              32%

                50%                                48%




          44%                           43%
                HSE area of                        HSE area of
                residence                          treatment
      Regional variation in % BCS use
                   (2000-04, 15-64, T1-2, not M1)
54.5%
overall




                    69%                                 71%
      39%                              38%

                57%                                 55%




          50%                           48%
                HSE area of                         HSE area of
                residence                           treatment
Regional variation in BCS use: model
BCS use within 6 months of diagnosis (final model*
  adjusted for age, TNM, grade, and method of presentation):
• significantly higher among patients resident in the
  Dublin/North-East administrative area (RR 1.16,
  unadj 1.18)
• significantly lower in the West (RR 0.73, unadj 0.64)
• not significantly different in the South (RR 0.98,
  unadj 0.87)
• compared with Dublin/Mid-Leinster
• Similar findings for area of treatment.

              *method of diagnosis and tumour morphology did not improve model fit
BCS variation by other patient and tumour
 characteristics: final area-of-residence model
 • significantly lower in age-groups 65-74 (RR 0.84) and
   75-99 (RR 0.91) v. 15-44
 • lower for T-category T2 (RR 0.72), T3 (RR 0.26), T4
   (RR 0.28) and T-unknown (RR 0.67) v. T1
 • lower for node-positive patients (RR 0.75), but higher
   for patients with unstated nodal status (RR 1.26) v. N0
 • lower for tumours of grade 2 (RR 0.94), grade 3-4
   (RR 0.90) or unknown grade (RR 0.86) v. grade 1
 • higher for screen-detected tumours (RR 1.25) or
   unknown presentation (RR 1.13) v. symptomatic
Hospital variation in % BCS use
          (2000-04, all ages, all stages)

                            Area of treatment

            DML             DNE         S       W      All
Centres   44.0% 64.1% 37.1% 29.1% 46.9%
              St James‟s    Beaumont   CUH      UCHG
             St Vincent‟s      Mater   WRH       LRH




Other acute 34.4% 43.9% 47.9% 32.9% 39.4%
general
Private   62.0% 71.3% 36.7% 33.3% 59.1%
Hospital variation in % BCS use
 (2000-04, ages 15-65, T1-T2, no distant metastases)

                             Area of treatment

              DML            DNE         S       W      All
Centres     50.8% 72.8% 45.1% 37.4% 56.2%
               St James‟s    Beaumont   CUH      UCHG
              St Vincent‟s      Mater   WRH       LRH




Other acute 39.7% 54.0% 53.3% 38.0% 45.4%
general
Private     66.7% 76.5% 38.1% 39.4% 64.2%
Hospital variation in BCS use: model
BCS use within 6 months of diagnosis (final model
  adjusted for age, TNM, grade, and method of presentation) :
• significantly lower in other acute general hospitals
  (RR 0.89, unadj 0.84)
• slightly but significantly higher in private
  hospitals (RR 1.06, unadj 1.26)
• compared with [what are now] designated specialist centres
            Regional variation in
        % radiotherapy use after BCS
                     (2000-04, all ages, all stages)

                        DML          DNE             S          W          All
    Area of           84.9% 88.8% 87.3% 84.4% 86.5%
    residence
    Area of           84.0% 89.1% 86.8% 85.7% 86.5%
    treatment

                                                    International comparison:
             80% (1985-2001) 32 studies, Malin et al. 2002 J Clin Oncol 20: 4381-93
c60% (1983) c75% (1996) US SEER stages I-II Nattinger et al. 2000 Lancet 356:1148-53
      Regional variation in
  % radiotherapy use after BCS
 (2000-04, ages 15-64, T1-T2, no distant metastases)

              DML      DNE         S       W       All
Area of     87.6% 92.7% 91.5% 89.8% 90.3%
residence
Area of     86.7% 92.3% 91.2% 92.4% 90.3%
treatment
Regional variation in radiotherapy use
          in BCS patients
Radiotherapy use within 12 months of diagnosis
  among BCS patients (final model adjusted for age,
  TNM, tumour morphology, and method of
  presentation) was:
• slightly but significantly higher among patients
  resident in the Dublin/North-East administrative
  area (RR 1.04, unadj 1.04)
• not significantly different in the West or South
• compared with Dublin/Mid-Leinster
• Similar findings for area of treatment
      Radiotherapy use (after BCS):
   variation by other patient and tumour
         characteristics: final model
• significantly lower in age-group 75-99 (RR 0.69) v. 15-44
• lower for T-category 4 (RR 0.85) & T-unknown (RR
  0.84) v. T1
• lower for unknown or unstated nodal status (RR 0.73)
  v. N0
• lower for “other adenocarcinomas” (RR 0.92) and
  unspecified morphologies (RR 0.42) v. breast-specific
  adenocarcinomas
• lower for incidentally detected cases (RR 0.85) v.
  symptomatic
Hospital variation in radiotherapy use
           in BCS patients
Radiotherapy use within 12 months of diagnosis
  among BCS patients (final model adjusted for age, TNM,
  tumour morphology, and method of presentation) was:
• not significantly different among patients treated
  in other acute general hospitals (RR 0.99) or private
  hospitals (RR 0.97)
• compared with designated specialist centres
 Survival equivalence of mastectomy
   and breast-conserving surgery?
• Assumed based on results of clinical trials
  comparing BCS plus adjuvant radiotherapy
  with mastectomy
• Broadly apparent in Irish data
• But not all Irish BCS cases have adjuvant
  radiotherapy (i.e. „full‟ breast-conservation
  therapy) thus some effect on long-term
  survival possible
under 65, all stages                                     age 65+, all stages         5-yr relative survival
                         5-yr relative survival

            0%     20%   40%      60%         80% 100%              0%         20%   40%      60%       80% 100%


                                     83.5%                  BCS only                      85.9%
    BCS only




   BCS + RT                           94.5%                BCS + RT                       96.9%




 mastect only                         85.7%              mastect only                      82.1%




mastect + RT                          78.3%              mastect + RT                      72.4%




                 5-year relative survival v. treatment:
                  all stages combined (2000-04) – all patients
under 65, all stages                                   age 65+, all stages         5-yr relative survival
                         5-yr relative survival

            0%     20%   40%      60%       80% 100%              0%         20%   40%      60%       80% 100%


                                      86.8%               BCS only                         93.7%
    BCS only




   BCS + RT                           94.7%              BCS + RT                          97.1%




 mastect only                         87.3%            mastect only                        86.3%




mastect + RT                          79.5%            mastect + RT                        74.7%




                 5-year relative survival v. treatment:
               all stages combined (2000-04) – 1-yr survivors only
under 65, T1-2(notM1)   5-yr relative survival        age 65+, T1-2(notM1)    5-yr relative survival

            0%    20%   40%      60%       80% 100%               0%    20%   40%      60%        80% 100%


    BCS only                          90.0%               BCS only                        96.5%




   BCS + RT                           95.5%              BCS + RT                         97.9%




 mastect only                         91.6%            mastect only                       89.0%




mastect + RT                          85.1%           mastect + RT                        81.8%




                 5-year relative survival v. treatment:
                 T1-T2 (not M1) (2000-04) – 1-yr survivors only
Relative survival modelling (2000-04 cases):
BCS + radiotherapy v BCS alone (1 5 yrs)           st



age   stage    adjustment                        EHR
15-64 T1-2     age, length of follow-up;          0.48
      (not M1) deaths<1 yr excluded          (0.26-0.89)
15-64 T1-2     age, T, N, length of follow-       0.53
      (not M1) up; deaths<1 yr excluded     (0.29-0.98)
15-64 T1-2     age, T, N, grade, chemo,            0.54
      (not M1) hormone, follow-up;           (0.29-0.99)
                  deaths<1 yr excl
                     Interpretation?
• Real influence on survival of missing radiotherapy? –
  unlikely of this magnitude? (but cf other pop v trial disparities)
• Confounded by other treatment (or lack of) e.g.
  chemotherapy? – likely(?); but addition of chemotherapy &
  hormone therapy to model changes little
• Confounded by institutional effects? – unquantifiable
  quality of treatment?
• Confounded by socioeconomic effects?
• Co-morbidity? – cause of death might or might not help
  clarify
• How many extra deaths each year resulting from
  under-use of adjuvant radiotherapy? – difficult
                   Further work
• Completeness of radiotherapy recording? (Private
    RT centres?)
•   Longer-term survival
•   Recurrence-free survival (better recurrence recording)
•   Appropriate use of axillary lymph-node sampling
•   Appropriate use of chemotherapy and hormonal
    therapy
                 Conclusions
 BCS use varies substantially between regions of
  residence or treatment and between hospitals
 Radiotherapy use after BCS varies less between
  regions/hospitals
 But 15% of patients having BCS for local or regional
  invasive breast cancer apparently do not receive
  adjuvant radiotherapy (45% of patients aged 75+).
 Across-the-board under-treatment of a minority of
  patients? – and slightly worse in some regions?
 Increased local recurrence and reduced longer-term
  survival expected; some tentative evidence of reduced
  survival in this study